D1.7.1 Data Format Specification

A CIP-PSP funded pilot action
Grant agreement n°325188
Deliverable
D1.7.1 – Data Format Specification
Work package
Due date
Submission date
Revision
Status of revision
WP 1, Requirements and Specifications
01/08/2013
30/07/2013
2.5
Responsible
partner
Contributors
DFN-CERT Services GmbH
Project Number
Project Acronym
Project Title
Start Date of
Project
CIP-ICT PSP-2012-6 / 325188
ACDC
Advanced Cyber Defence Centre
01/02/2013
KU Leuven
TU Delft
Various project partners for the questionnaires
Dissemination Level
PU: Public
PP: Restricted to other programme participants (including the Commission)
RE: Restricted to a group specified by the consortium (including the
Commission)
CO: Confidential, only for members of the consortium (including the
Commission)
X
Version history
Rev.
Date
Author
Notes
1.0
30/05/2013
DFN-CERT
TOC
2.0
12/07/2013
DFN-CERT
Initial version of content
2.1
16/07/2013
DFN-CERT
Internal comments addressed
2.2
16/07/2013
DFN-CERT
Internal comments addressed
2.3
16/07/2013
TU-DELFT
Contents for section 3.4
added
2.4
18/07/2013
DFN-CERT
Added Introduction and
Summary
2.5
29/07/2013
DFN-CERT
Layout improvements
Glossary
ACDC
CCH
Advanced Cyber Defence Centre
Centralised Data Clearing House
Table of contents
1.Executive summary..............................................................................................................4
2.Introduction........................................................................................................................... 4
2.1.Structure of the Document.............................................................................................5
3.Data Formats and their Requirements..................................................................................6
3.1.User Groups................................................................................................................... 6
3.1.1.End Users................................................................................................................ 6
3.1.2. Internet Service Providers (ISPs) ...........................................................................6
3.1.3.Intrusion Detection Systems (IDSs) ........................................................................7
3.1.4.Analysts ..................................................................................................................7
3.1.5.Computer Security Incident Response Teams (CSIRTs) .........................................7
3.1.6.CCH operator ..........................................................................................................8
3.1.7.Vendors ...................................................................................................................8
3.1.8.Law Enforcement Agencies .....................................................................................8
3.1.9.Anti Botnet Initiatives ...............................................................................................8
3.1.10.Private Hosting Companies ...................................................................................8
3.1.11.Industrial Users .....................................................................................................9
3.1.12. Press and Media ..................................................................................................9
3.2.Technical Requirements ................................................................................................9
3.2.1.Set of Recommended Criteria .................................................................................9
3.2.2.Set of Additional Criteria .......................................................................................12
3.3.Organizational Requirements ......................................................................................12
3.3.1.Set of Recommended Criteria ...............................................................................12
3.3.2.Set of Additional Criteria .......................................................................................14
3.4. Content Requirements ...............................................................................................14
3.5.Legal Requirements ....................................................................................................15
3.5.1.Legitimacy of Processing.......................................................................................15
3.5.2.Data accuracy and Finality.....................................................................................15
3.5.3.Proportionality........................................................................................................15
3.5.4.Confidentiality and Security....................................................................................16
3.5.5.Transparency......................................................................................................... 16
2
Page 2 / 95
3.5.6.Conclusion of Legal Requirements........................................................................16
3.6.Conclusion of Requirements........................................................................................16
4.Stocktaking of Relevant Data Formats...............................................................................16
4.1.Binary Data Formats....................................................................................................17
4.1.1.IPFIX/NetFlow and sFlow.......................................................................................17
4.1.2.Hpfeeds................................................................................................................. 17
4.2.Textual Data Formats...................................................................................................18
4.2.1.XML....................................................................................................................... 18
4.2.2.JSON..................................................................................................................... 20
5.Evaluation of the Data Formats..........................................................................................21
5.1.Evaluation of the questionnaires..................................................................................22
5.1.1.Overview of the received questionnaires and referred data formats.......................22
5.1.2.Overview of the technical properties of the data formats........................................23
5.1.3.User groups and their requirements.......................................................................24
5.1.4. Overview of data formats and supported sources ................................................25
5.1.5.Use Cases............................................................................................................. 26
5.2.Summary of Results.....................................................................................................27
6.Conclusion.......................................................................................................................... 28
7.References......................................................................................................................... 29
8.Appendix............................................................................................................................ 30
8.1.Questionnaire............................................................................................................... 30
8.2.Questionnaire A........................................................................................................... 32
8.3.Questionnaire B........................................................................................................... 35
8.4.Questionnaire C........................................................................................................... 38
8.5.Questionnaire D........................................................................................................... 43
8.6.Questionnaire E........................................................................................................... 47
8.7.Questionnaire F............................................................................................................ 51
8.8.Questionnaire G........................................................................................................... 60
8.9.Questionnaire H........................................................................................................... 64
8.10.Questionnaire I........................................................................................................... 68
8.11.Questionnaire J.......................................................................................................... 70
8.12.Questionnaire K......................................................................................................... 73
8.13.Questionnaire L.......................................................................................................... 75
8.14.Questionnaire M......................................................................................................... 77
8.15.Questionnaire N......................................................................................................... 81
8.16.Questionnaire O......................................................................................................... 83
8.17.Questionnaire P......................................................................................................... 87
8.18.Questionnaire Q......................................................................................................... 92
Table of figures
Fig. 1: Sources for data distribution into the Centralised Data Clearing House .......................5
Fig. 2: Overview of the received questionnaires....................................................................23
Fig. 3: Properties of data formats..........................................................................................24
Fig. 4: Number of formats supporting a specific user group...................................................25
Fig. 5: Number of data formats supporting a specific user group...........................................26
3
Page 3 / 95
D1.7.1 – Data Format Specification
1. Executive summary
The aim of the ACDC project is to set up a European Advanced Cyber Defence Centre
(ACDC) to fight botnets. To reach this goal, the project will introduce components and
workflows to gather and analyse data originating from technical sensors such as honeypots
and IDSs as well as user reports. A central role in ACDC is devoted to the centralised data
clearing house (CCH) providing a platform for storage and analysis of gathered data. It is
important to note, that the data does not solely vary by technical sources, but also by
different user groups that are involved. Since each user group and technical sensor does
have different requirements the choice of applicable data transport formats is a crucial task.
In this document the relevant data formats are assessed, and a first advice is given on which
formats are expected to be important for the later usage in the context of the ACDC project.
The first necessary step is to assess the available data formats and their properties. This is
done by analysing a survey in collaboration with all ACDC partners. The survey is based on a
questionnaire that has been distributed to the partners comprising questions about the data
formats in use, their use cases, properties, and planned extensions. This survey reveals a list
of 15 different formats from which nearly all are based on a textual representation of the data.
Because of their formal structure 10 formats allow an automatic processing of the data
whereas 7 formats provide a publicly available specification of its syntax and semantics. The
responses include some well-known formats such as IODEF, X-ARF, and STIX. Other
applied formats are devoted to a specific use case like the sFlow format for transferring
NetFlow data and hpfeeds, which is specific for honeypot data.
In a second step we assess a list of technical, organisational, and legal requirements that
have been derived from publicly available specifications of data formats as well as a
preliminary analysis of the projects' planned evaluation tasks and the legal framework. In
addition, a list of use cases and their specific requirements has been compiled that are
relevant for the project. It turns out that the available data formats meet these requirements
pretty well. However, all data formats lack a fine granular specification for restrictions
concerning the usage of included data. This shortcoming could be either compensated by an
extension of the data formats itself or by considering this feature in the transport protocol.
2. Introduction
The intention of the ACDC project is to set up a European Advanced Cyber Defence Centre
(ACDC) fighting botnets. ACDC's approach is to
• foster an extensive sharing of information across borders to improve the early detection of
botnets
• provide an extensive set of solutions accessible online for mitigating ongoing attacks
• use the pool of knowledge to create best practices that support organisations in raising
their cyber-protection level
• create a European wide network of cyber-defence centres
ACDC will deploy a comprehensive set of national support centres throughout eight Member
States interconnected to ACDC's central clearing house (CCH). Through this networked
approach, ACDC will also pave the way for a consolidated approach to protect organisations
from cyber-threats and support mitigation of on-going attacks through easy access to an
increasing pool of solutions.
4
Page 4 / 95
A central role in ACDC is devoted to the centralised data clearing house. This component
collects all data gathered by technical sensors such as honeypots and IDSs as well as user
reports. As a fundamental advantage, the CCH provides a central platform to analyse and
process the data allowing to completely reveal botnets and other global incidents by attack
data correlation and to distribute the resulting enriched data.
As shown in Fig. 1 the CCH collects data from various sources that comprise technical
sensors as well as user reports. It is important to note, that the data does not solely vary by
technical sources, but also by different user groups that are involved. All this results in
different requirements regarding the data exchange and processing. For example, some
technical sensors produce large amounts of attack data requiring an efficient way for their
submission. Some user groups might contribute data intended for reporting security incidents
and supporting research. While reporting incidents could not be done without exact
information, legal restrictions might require an anonymisation for any other usage of the data.
It is reasonable to assume, that all these requirements cannot be fulfilled by a single data
exchange format. Instead, a bunch of formats is needed contributing the different properties
as required. This document aims to collect a list of formats that are already in use in the
ACDC community and to investigate whether these formats meet the demands of identified
use cases. An additional aim of this document is to identify shortcomings resulting from the
project demands not addressed by the data formats already in use and to give advices how
to close these gaps.
Fig. 1: Sources for data distribution into the Centralised Data Clearing House
2.1. Structure of the Document
The document is structured as follows: Section 3 gives an overview of identified technical,
organisational and legal requirements to be used for evaluating data exchange formats. In
the following Section 4 we describe the properties of data formats resulting from a survey
wherein the partners contributed information about the data formats in use. Therefore, these
formats can be expected to be relevant for the ACDC project. Section 5 gives an overview of
the conducted survey. To decide which data formats are applicable we summarise relevant
use cases in this section and enumerate their fundamental requirements that must be met by
5
Page 5 / 95
a data format. Considering these requirements applicable data formats are listed for each
use case. In addition, we identify missing demands not addressed by data formats already in
use and propose how to fill those gaps. The results of the survey are shown in the Appendix.
3. Data Formats and their Requirements
In this section common requirements for data exchange formats are specified and explained.
We split the content of this section into five different categories of requirements: user groups,
technical requirements, operational requirements, content requirements and legal
requirements. The presentation of criteria is concluded in a final section.
The requirements are listed as presented. There is no particular order of relevance, neither of
nor within the categories themselves. Some requirements can also be implemented by the
communication protocol used for the data exchange and therefore discarded for the data
exchange format. Since there is no specific communication protocol or requirements for the
communications protocol to be used these requirements apply to the data exchange in
general and might therefore be handled by the data exchange format.
3.1. User Groups
This section lists the user groups participating in the ACDC data exchange. Each
participating user or system may have a different set of requirements depending on its
involvement, data exchange scope and communication amount with the CCH.
This list is by no means complete and not all listed user groups will actually transmit or
receive data from the CCH. For a more detailed description of each user group see ACDC
WP6.
3.1.1. End Users
For all users their privacy must be respected and protected. Therefore legal requirements as
stated in Section 3.5 must be observed. Anonymization is as a best practice implemented by
the reporting tool itself as stated in Section 1.1.6.1 in ACDC's Description of Work on Page
31.
Whereas in other cases users want their personal data be handled for involvement or
notification purposes. In this case the appropriation of the provided personal data must be
observed. Most likely a user wants an immediate classification or result of the malware's
analysis she reported. If this is not feasible she might want to be informed when a manual
analysis is completed.
Also for information purposes the end user is to be considered a receiver of data from the
CCH, especially the results of analyses. This also includes the public domain unless access
to this information is to be restricted.
3.1.2. Internet Service Providers (ISPs)
ISPs can submit data on malicious traffic pattern to the CCH. On the other hand the ISP
should contact their customer to remove malware, maybe using the ACDC's project web
page.
ISPs might also be informed of malicious hosts within their network to contact the customer
to sanitize their infected hosts. Since all ISPs provide abuse contact details this data can be
used to report detected malicious hosts. Therefore ISPs contacted in this matter do not have
to be a contributing partner to this project.
6
Page 6 / 95
3.1.3. Intrusion Detection Systems (IDSs)
The information gained from analysis can be used to define new intrusion and attack
schemes for intrusion detection and help in discovering weaknesses used for an attack.
Therefore intrusion related data could be shared with the CCH, but will most likely have to be
cleaned from identifying data. Depending on the amount of data this additional analysis might
not be manageable and therefore not uploaded to the CCH. But this data could be
extraordinarily helpful.
IDS/IPS can also be updated using recently discovered malware (or malware behaviour).
Therefore IDSs/IPSs should also be considered receivers of analysed and classified data
from the clearing house. On the other hand, this update process will most likely not be
automated, since system administrators will not simply trust traffic considered illegitimate by
someone else. The update procedure might involve the vendor of the IDSs/IPSs.
Drones
Information security (ISEC) specialists also use drones, infected hosts receiving commands
from the botnet's c&c server but not executing them. With these drones they are able to
record the commands sent out to the botnet's bots. The captured traffic pattern may then be
used to identify botnets in legal network traffic. The result may be shared with the CCH.
Honeypots
High-interaction honeypots are vulnerable hosts placed to be infected by malware. Most
often the operating system is running on a virtual machine, so the operator is able to trace
the operations used by the malware during infection and execution. Using a virtual machine
she is also able to take memory snapshots and have a special environment (called sandbox)
set up from which the malware cannot infect other hosts but at the same time execute the
commands of the c&c server without any effect. At that stage the infected host in the
honeypot works as a drone. Additionally, low-interaction honeypots are used to detect attacks
and to capture malware. In contrast to a high-interaction honeypot, this type only simulates a
vulnerable service.
3.1.4. Analysts
The ACDC architecture incorporates analysts working with the data stored in the CCH. They
—mostly automated―analyse the provided malware or reports and classify them. But they
will most likely also perform manual analysis of malware or reports that cannot be classified
by software. On the one hand these analysts take data samples from the CCH but they also
provide further data or information to be stored in the CCH or they enhance reports with
results of their analyses.
The group of analysts also includes academic and business researchers as well as ISEC
specialists and employees of Computer Security Incident Response Teams (CSIRT).
3.1.5. Computer Security Incident Response Teams (CSIRTs)
CSIRTs (or CERTs) are teams of IT security specialists dealing with incident handling and
Information Security Management. They are highly interested in up to date data on malware
spreading and recently upcoming malware.
While CSIRTs that are larger, or more successful in networking already have their channels
receiving information about malware infections shortly after discovery, especially smaller or
more isolated CSIRTs might be interested in data exchange.
Depending on their contractual situation CSIRTs might not be able to provide data on
malware infections but are interested in receiving data of active malware.
7
Page 7 / 95
3.1.6. CCH operator
The operator of the Centralized Clearing House will be able to access all incoming and
outgoing data. So organizational controls must be implemented to establish access control.
As the operator of the CCH service does not provide any transmitting or receiving of malware
reports other than those required for the operation of the service the operator is included in
this enumeration for completeness only.
3.1.7. Vendors
Vendors of Anti-Virus (AV) or Firewall (FW) software or appliances are interested in malware
as well. They are interested in malware samples as well as the resulting analysis to close
vulnerabilities in their products. And also vendors of Operating Systems (OS) are interested
in exploitation data too.
On the other hand these vendors might not be willing to share vulnerabilities or weaknesses
of their products.
3.1.8. Law Enforcement Agencies
Law enforcement agencies (LEA) might want to access reported data in case of ongoing
investigations or by court order. The question whether there are legal requirements or even
an obligation to share data with LEAs is covered in more detail in Deliverable D1.8.2 of the
ACDC project.
3.1.9. Anti Botnet Initiatives
Anti Botnet Initiatives like the ACDC consortium might be interested in data exchange as
well. These include national anti botnet initiatives as well as national anti botnet advisory
centres which are not part of the ACDC consortium.
3.1.10.Private Hosting Companies
This includes Website operators, hosting companies, data centres and domain providers.
Website operators must be notified to clean and stop it from spreading malware any further
when their Content Management System (CMS) is infected. To contact the website operator
there must be abuse contact details available. If not, the hosting company must have
established an abuse contact team to handle the take down notice or cleaning request. This
abuse team has to contact the website owner or take down the website themselves
depending on the severity and the contractual situation.
Hosting companies might often be a faster replying and more reliable point of contact when
trying to take down an infected Internet service. While they are scanning the network traffic
data centre operators do want to prevent damage and illegal usage of its infrastructure.
Therefore also data centres might be able to share their findings of illegitimate traffic with the
CCH. On the other hand data centres might not be willing to share their customers' traffic
with the public. Data centre operators as well as hosting companies are most likely operating
IDSs and IPSs to protect their infrastructure and their customers' data.
Domain providers might be willing to take down fast-flux domains, for which the domain
serving IP address is changed frequently. These domains are most likely used for malware
distribution because it's more complicated to predict the next IP address the domain points to
in future. Therefore the easiest way to take down the malware distribution centre is by
shutting down the domain and its Domain Name System (DNS).
8
Page 8 / 95
3.1.11. Industrial Users
System administrators might be interested in receiving data about recent malware spreads to
update their heuristics and IDSs/IPSs but are not likely to share malware infections with
someone outside of the company unless required by law.
This is especially valid for enterprises operating critical infrastructure or banking companies
since these might be even less willing to share security related infections resulting in
negative publicity and therefore loss of customer trust.
So we assume this category might only be interested in receiving data. On the other hand
small and medium enterprises (SMEs) are more dependent on someone else's IT
infrastructure or security related services.
3.1.12.Press and Media
Press and Media related services play an important part in today's society. They uncover
incidents and investigate cases brought up by notifications and information provided by
whistle-blowers. Therefore press and media might be a provider of data for the CCH and on
the other hand a receiver of statistical data or exemplary cases.
In other cases the press must be used by companies dealing with personal data to inform the
public on issues of loss of data when other means are not appropriate or the number of
people affected by a loss of data is so great that it'd not be efficient or possible to notify each
person on its own.
3.2. Technical Requirements
Technical requirements covering details needed by machines to work with the provided data.
We split the criteria into a set of recommended criteria and additional or optional criteria.
3.2.1. Set of Recommended Criteria
Machine Readable Data Format
As with all data exchanged by computers the data transmitted must be structured. For this,
structural elements must added to the content so machines can separate the data fields.
Those structural elements must be used according to a defining standard for the data
format's language. See Criterion "Text-Based Data Format" for more details on this matter.
This criterion is obvious and must be met, since reports written in human language might not
be parse-able for machines and therefore not understandable. And as the amount of data to
be transmitted cannot be handled manually the format should be machine readable.
The data format should be validated whether it is the "right thing". For this, a formal validation
is to be carried out to decide whether the data format and its data fields are sufficient for the
intended purpose.
Text-Based Data Format
In general, we recommend a text-based data format for reporting attacks and incidents since
these ease the encoding and data format issues handled later on in this report. During
development text-based data formats are more suited for bug tracking and for creating
erroneous situations, falsified messages that are not compliant to the standard.
Also the focus of exchanged data is on text-based components, but as the text elements
have different data field lengths the advantage of binary formats (directly accessible data
fields due to offsets) diminishes. Since encoding standards for binary data in text-based
formats exists (e.g. Base64 encoding) it is possible to transmit binary data like malware
9
Page 9 / 95
samples or screenshots within the data exchange format. It is important to select a data
format that supports attachments (even though this is a more content related requirement).
A text-based data format is not the best format for storage of the data since for iteration over
the reports a database-based solution is much more efficient. But as the data format in this
case is only for exchange of data the less efficient storage format is not important.
Text-based data formats are much easier to extend as existing parsers can ignore the
additional parts. But changes to the data format become backwards incompatible if
mandatory parts in a previous version are not available in a more recent one. A text-based
data format is usually defined by a Document Type Definition (DTD, for XML), defining the
legitimate usage of blocks and elements. These can be used by parsers and serializers to
verify their input and output, respectively. Those definitions can also be used to generate
parsers automatically without any implementation issues.
Even though a text-based data format is slower to process (parsers for binary data can use
fixed byte positions for faster access to selected data fields) this is likely only done once
upon receiving the message. Despite the complexity developing a parser for text-based
formats messages in this case are rather simple and the data formats in use are already well
known, so robust parsers most likely already exist.
Text-based formats are platform-independent, whereas binary encoded messages have to
define the range of numbers or how data is to be interpreted by the parser (endianness of
word/integers).
Internationalization
The data format should support internationalization (i18n) and localization (l10n) since this is
a European and therefore multi-language project and not all end users will be able to state
their report using the English language. So the data format must be able to transmit
characters of all European languages (specified by the encoding of the data file, e.g. UTF-8).
The receiver of the file must be notified of the file's encoding upon transmission, to use the
right encoding for decoding. Otherwise characters might be missing or displayed wrong in the
text possibly leading to false data.
Next to the encoding issue is the criterion that reports might be multilingual whereas some
parts might be in one and other parts in another or several other languages.
Ensuring Security and Message Safety
The message's security and safety could be handled by the underlying communication
protocol but if that protocol may not be able to guarantee these factors they can be handled
by the data format itself.
The data message must have some data fields to verify the message's integrity. This is
usually done with (cryptographic) checksums. The message's integrity is saved if the
message was not altered since checksum creation.
For some message contents protecting the confidentiality during transmission might be
valuable. Therefore the content of the message is encrypted, leaving the problem of
encryption key distribution. Usually a public key infrastructure is used but in terms of criteria
for the data exchange format it is merely important whether the data format supports
encryption if the underlying communication protocol does not.
It would be wise if the sending and receiving peers were able to verify the other peer's
identity. This could be done with a public key infrastructure and encryption. Usually the
authentication is handled on communication protocol layer.
10
Page 10 / 95
Another important task is to prevent message duplication since this could lead to denial of
the service. As this is usually established on protocol layer other means must be used to
prevent a message being send several times. This could be a unique message id or a limited
timeslot for which the message is valid.
Documentation Of Data Format
To prevent misunderstandings due to different interpretations of the provided data the
documentation of the data format must be up to date and the specification of the data format
must be unambiguous.
This is important for each data type and each data field. E.g. whether the string value of 'true'
in the transmitted text-based message is interpreted as the boolean value of true, the string
value of 'true' or as the integer value 1.
Especially important is the format of timestamps. These data fields should incorporate the
time zone of the host. Usually this is not a problem when timestamps are formatted as strings
but it turns into a problem when timestamps are transmitted in seconds since a specific date.
Supporting Bulk Messages
If possible the data format should support the transmission of several messages in one bulk
message. Therefore the data format may act as a container comprising messages.
The receiver must decide whether to separate the messages into several reports or one large
report. There could be a connection between several findings on one machine, which is a
piece of information that would get lost if the message is split (See criterion in Section 3.2.2
under “Link between reports”).
Supporting IPv4 and IPv6
If data formats use data fields for IP addresses the data format specification should be able
to handle IP addresses of version 4 as well as IP addresses of version 6 for future use and
long usability.
As an IP address date must be considered as possibly human related there must be means
to anonymize the field and indicate the anonymization to analysts.
Vendor Independence
The data format specification should be free of charge and defined in a free and open format.
The reason is to not become dependent of the good will of the vendor or the software
solution.
Using an open format allows to exchange the used software solution if the software is
abandoned, or not sufficient, or not compliant to the projects' goals any more. An open format
also allows to extend and adopt the data format for the project's requirements and allows to
implement parsers for all platforms, especially for new, emerging platforms possibly in
competition with a vendor.
Version Tag
A data format needs a version tag to support extendibility, whereas a parser for a later
version should be able to read a report formatted in a prior version of the data format. The
version tag can also be used to identify backward and forward incompatible changes in the
data format specification.
A version information should be standard in every data format specification as it is required
for further improvement and development.
11
Page 11 / 95
3.2.2. Set of Additional Criteria
There are two criteria which are merely optional: Using an object-oriented data format and
support of compression.
Object-Oriented Data Format
As the object-oriented notation is the natural way of describing items it might be wise to
select a data format supporting this notation. The object-oriented notation includes attributes
for objects and also nesting of objects. The applicability of object-orientation depends on the
structure of the reports to transmit. For simple reports or data structures an object-oriented
orientation is overload.
Support for Compression of Content
The content of a report could be compressed to save bandwidth during transmission. While
executing compression on mobile devices might reduce the battery by a larger amount than
saved by less data transfer. So compression should be optional in terms of using it if the data
format provides it.
It should be considered, that compressed data must be encoded text-based when
transmitted in a text-based format. Therefore the compression might not save that many
bytes since the encoding adds bytes (e.g. see Base64-Encoding).
3.3. Organizational Requirements
Another set of criteria to evaluate diverse data formats are organizational requirements.
These deal with individual requirements raised by users on usage of their provided data,
support of the ACDC workflow and licensing issues. As done with the technical requirements
these are split into recommended and optional criteria.
3.3.1. Set of Recommended Criteria
Anonymization
As required by law in some countries (e.g. Germany) all data relating to a person must be
anonymized. Exceptions for required anonymization of data are the user's consent whereas
the consent must be free of choice, or if there is a law requiring the data. The last exception
occurs if there is a legal binding between the user and the storage unit.
While this is easy to handle for the submitting user it's much harder to achieve for the
attacking or third party. Please see the legal requirements section in Section 3.5 for further
details and deliverable [D1.8.1].
Well-Defined Syntax and Semantics
As already stated in Section 3.2.1 under “Documentation of the data format” the data format
must be well-defined for unambiguity. That includes well defined syntax and well defined
semantics. Whereas syntax defines the construction of valid documents, semantics define
the meaning of elements and how to interpret them.
Having well defined syntax and semantics allows validation of the data format (Whether the
data format allows all required data to be transmitted) and automated verification of reports
based on the data format (Whether the reports are legitimate based on the data format
specification).
12
Page 12 / 95
Individual Requirements
As individual requirements on the usage of provided data might arise it would be a wise
choice if the data format supports the specification of individual requirements concerning the
data usage. Therefore the data fields or alternatively the protocol to submit data should
provide attributes allowing to state requirements or restrictions using some formal language.
Please see ACDC's Deliverable [D6.1.1] for an analyst's point of view on this topic. Each
partner who requests to analyse data has to define what he intends to do with the data to
receive. Those intended purposes could then be selected by the submitting user.
There should also be a possibility of applying timing constraints, which are executed after a
defined period of time. Those might include anonymization or deletion of provided data.
There might also be a legal requirement to anonymize or delete data after a period of time
defined by law or court order (e.g. seven days in Germany for logging data used to detect
malicious behaviour or fraud). These timing constraints could be set by user or by law.
Confidentiality
Additionally to technical requirements of message safety and content security there exists an
organization requirement to ensure that only authorized users are able to view the provided
data. This means to employ access restrictions to the message's content on all level either by
encryption of the communication protocol and in the software environment at the CCH.
Data fields are needed in the data format to support specification of confidentiality restriction.
Appropriation of Human Related Data
The data format shall ensure that provided data is only used for the intended purposes stated
unambiguously and well-defined during survey. So called appropriation must be ensured at
all levels at the transmission and storage at the CCH and during evaluation. This is also a
legal requirement introduced by law.
All additional purposes are generally forbidden by law. But narrow exceptions do exist.
Support of the ACDC Workflow
Where applicable the data format shall support the ACDC workflow, meaning that data fields
should match required data fields for analysis or even finer but not more coarsely so the
analysing software would have to split the data manually.
Providing the data in the right data fields supports automated analysis.
Extendable Data Message
The data format shall be extendable. That includes data fields (e.g. for additional free text)
defined by a data format extension. These might also include vendor specific extensions to a
data message (e.g. name and version of the software creating the report).
Therefore the specification should not be too tightly tied down, but to allow individual
improvements or adjustments. Additionally the parsers must be adopted to the new format.
Depending on the specification actually used, the parsers are generated automatically.
Licensing Issues
As already stated above in Section 3.2.1 under “Vendor Independence” the data format
should be in a free and open format to be free of charge. Especially after the initially project
phase as a pilot project and funding by the EU costs should be kept down.
But not only the costs are dependent on the license, also the usage of the data format might
be limited by the publisher/owner of the data format. Therefore a free specification is the best
13
Page 13 / 95
option, especially when the data format is to be extended. This might not be allowed in all
licenses especially when the license is proprietary.
3.3.2. Set of Additional Criteria
Restrictions on Subsets of Data
If possible the data format should allow to define restrictions only to subsets of data in a
report. Therefore the data format should be flexible to add restrictions to any subset of data.
These may include timing constraints as stated in Section 3.2.1 under “Individual
Requirements”.
This criterion is marked as additionally since most format will not support it.
Stating Confidence in Report
To prioritize manual analysis of important reports the data format might be able to allow the
user to state the confidence into the finding and the severity of the incident. On one hand this
would result in an enhanced user involvement but could on the other hand lead to
inexperienced users stating the problem as severe whereupon the problem is merely an
annoyance.
Therefore the user's profile could be equipped with a credibility score determining the
experience of the user. But this would require users to register with the service and lead to
problems as how to deal with first time submitting users being ITSEC professionals.
Linking Related Reports
The data format could allow links to related reports happened before or at the same time
(see also Section “Supporting Bulk Messages”) to establish relationships between reports.
The connection between related reports could be established manually or by automated
analysis.
The intention of establishing connections between reports is to create a data-warehouse
dealing with 'big data' to gain even more information from reports.
3.4. Content Requirements
As discussed in Section 3.3.1 and in [D1.2.1], input data formats supported by CCH ought to
guarantee anonymity and privacy. However, we should not neglect the impact of these
security requirements in the quality of the analysis in WP4 and in other work packages. For
example, consider anonymizing IP addresses using a certain mathematical function (e.g.,
hash function). While we will be able to tell how IP addresses' measure of evilness changes
over time, we will not be able to tell what addresses they are and, consequently, will not be
able to use the output in real-time IDS.
In this sense, we should evaluate each data type on a case-by-case basis, i.e., what fields
must be anonymized to keep both user's anonymity and privacy while, at the same time,
keeping the highest quality of analysis possible.
For example, consider a spam message. In this case, one could consider to remove both
sender and destination e-mail addresses from the contents, and/or maybe entirely the
content the message, keeping only the metadata (e.g., source/destination IP addresses,
timestamp, etc.). However, to keep the user's privacy in the metadata, one could remove the
last octet of the IP addresses (e.g., 192.168.0.x instead of 192.168.0.53). Another example
is the case of DDoS attacks reported using IPFIX/NetFlow [RFC 5101]. For this case, the
report will only list the metadata, and not the message contents associated with the attack.
Taking into account the heterogeneity among data sources and exploited applications, we
recommend:
14
Page 14 / 95
1. Define unique anonymization functions for fields (or a single function).
The functions must be consistently used across all datasets to enable correlation between
various data sets.
2. For each type of data and format, evaluate which fields compromise both anonymity and
privacy (or fields that the contributor requires to be anonymized).
3. Then, determine if they must be anonymized. If yes, then employ functions defined in 1.
4. Evaluate the results to ensure privacy and anonymity.
It is important to emphasize that this task should be executed in an interdisciplinary fashion,
considering both technical and legal requirements.
3.5. Legal Requirements
This section is a short summary of the D 1.8.1/2 – Legal Requirements, which will clarify the
rules applicable to the project and in particular to the mitigation/detection tools to be
deployed. The processing of personal data (any information relating to an identified or
identifiable natural person, e.g. IP address, email address, etc.), requires observation of
stringent protection rules. As result, partners involved in this processing must comply with the
principles of legitimacy, data accuracy and finality, proportionality, confidentiality and security,
and transparency.
3.5.1. Legitimacy of Processing
The legitimacy of processing lies on the unambiguous, specific, freely given and informed
consent of the data subject (person to whom the data relates). In principle, the partner with
whom the end user has a direct contractual relationship (or is subject to in the case of a
public mandate) is best placed to register user’s consent as far as the collection or release of
her/his personal information is concerned.
3.5.2. Data accuracy and Finality
The data accuracy requirement entails the data controller obligation of putting in place
mechanisms and procedures that ensure the reliability of the personal data he/she
processes. Furthermore, the principle of finality or purpose limitation dictates that the
authorized usage of personal is restricted to the specified, explicit and legitimate purposes for
which it was first collected.
3.5.3. Proportionality
Data processing also needs to be proportional, implying that:
1. There must be a sufficiently narrow correlation between the (legitimate) purpose
articulated by the controller(s) and the data being collected;
2. Personal data should only be disclosed or otherwise made available to the extent that it is
necessary to achieve the purposes of the processing;
3. Personal data should not be maintained longer than is necessary for the purposes for
which the data were collected and/or further processed;
4. Controllers should seek to minimize the number of copies of personal data being
processed;
5. If the purposes of the processing can also be realized by less intrusive means, i.e. by
means which are less likely to have an adverse impact on the privacy or other
fundamental freedoms of the data subject, such means should be used;
15
Page 15 / 95
6. Even if legitimate, the processing may not prejudice the data subject in a way that is
disproportionate in relation to the interests pursued by the controller.
3.5.4. Confidentiality and Security
Obliges data controller to implement appropriate technical and organizational measures to
ensure the confidentiality and security to protect personal data against accidental or unlawful
destruction or accidental loss, alteration, unauthorized disclosure or access.
3.5.5. Transparency
The transparency principle gives data subjects the right of being notified of the processing of
her personal data (notice), having means to obtain further information (right of access) and
immediate tools of recourse towards the controller in case she feels her data are being
processed improperly (right to rectification, erasure or blocking).
3.5.6. Conclusion of Legal Requirements
Fewer legal restrictions apply to data that cannot be related to an identified or identifiable
person, as they are out of the scope of data protection regulation. Overall, it means that nonpersonal or anonymized data require less legal consideration according to their usage and
can be processed in a simpler way. Whenever the relation to a person is not required,
partners are asked to convert the processed personal data into anonymized data, a form
which does not identify individuals and does not allow re-identification through data matching.
Anonymized data shall be preferred whenever it does not significantly harm the outcome of
the mitigation/detection tools.
3.6. Conclusion of Requirements
The technical and organizational requirements are merely sets of criteria that should be met
by the data formats already in use. If a data format is to be selected from scratch these
criteria can be seen as an evaluation catalogue on how to find the most suitable data format.
Whereas not all criteria have to be met. For some there are workarounds by implementing
the requirement at communication protocol layer.
While technical and organizational requirements are basically defined by technicians
implementing the data exchange, content and legal requirements are defined by analysts and
the legal situation, respectively. Whereas analysts try to get as much data as possible for the
analysis, the legal department determines what data can be obtained and which processing
(e.g. anonymization or pseudonymization) can be done with the data depending on the
processing's legal background. These are two very diverse vantage points and the result—if
and to which extend each data field could be used for analysis—will be somewhere in
between. With anonymized data analysis is hardly feasible (or the results lack quality or
significance), but breaking the law is no option either. So it has to be defined—as expressed
in Section 3.4 in the second paragraph and Section 3.5.6, respectively—to which extend data
must be anonymized to be legitimately used for processing at all. This results in a trade-off
between what is allowed and what is required, obeying the risk mitigation strategy in
Section 2.4.3 in [DoW].
4. Stocktaking of Relevant Data Formats
This section gives an introduction to the relevant data formats. Because of the massive
number of formats we focus on a selection resulting from a survey by which a questionnaire
was distributed among the ACDC partners. The data formats can be structured using
different choices of criteria. In this section we classify the formats according to the encoding
which can be either binary or textual. A binary encoded data format can, for example, be
related to a structure in the programming language C. Thus, the data records are structured
according to a C variable structure. Textual formats use the character encoding such as
16
Page 16 / 95
ASCII or UTF-8. They can be either formally structured by using XML or lack a structure at
all.
4.1. Binary Data Formats
As previously mentioned binary encoded data formats are typically related to a structure in
the programming language C or a similar language. Their most important advantage is the
compactness of the messages, because there is no need to use textual metadata separating
data fields, as used, for example, in XML. However, they require adequate computer
programs for their processing. In general, binary data formats are advantageous to transfer
large amounts of bulk data.
4.1.1. IPFIX/NetFlow and sFlow
IPFIX/NetFlow [RFC 5101] and sFlow are part of a family of protocols to transfer metadata
related to the fundamental information of network connections (NetFlow) which include the
following information:
• IP addresses
• Port numbers
• TCP/IP Flags
• Number of packets in the flow
• Size of the transferred data.
The IPFIX/NetFlow protocols are initially supported by network equipment such as routers to
export data of monitored network connections. Therefore, the primary intention of this
protocol family is to gather and transfer NetFlow data on routers that may include productive
traffic of a large network.
NetFlow data support the detection as well as forensic analysis of computer security
incidents. This data is very valuable to react to distributed denial of service (DDoS) attacks.
In addition, a large number of security tools analyse NetFlow data for anomalies that may be
caused by large scale attacks and technical problems. Another important use case is to
reveal the full extent of an incident in a forensic investigation. NetFlow data allows to track
connections that are originated by a compromised system. This is especially important to
track botnets because this allows to monitor network connections that either target or
originate from a control and command server. Botnets are either controlled by a central
server or a peer-to-peer network structure. In the first case, NetFlow data allows to track
down other systems that connect to this server. These systems are likely also compromised
and controlled by the server. In the second case, NetFlows could help to track the peer-topeer structure of the botnet. Therefore, NetFlows play an important role in the tracking and
investigation of botnets.
4.1.2. Hpfeeds
Hpfeeds is a data exchange format especially dedicated to the exchange of honeypot data.
Currently, some honeypots including Dionaea and Glastopf natively support the protocol. The
idea behind hpfeeds is to supply and receive honeypot data on “data feeds”. Feeds are
implemented as a bus-like architecture and are isolated by different channels. Thus,
honeypots supply data to a central bus where receivers can subscribe to data they are
interested in. Optionally, the channel can be secured using SSL/TLS. A receiver has to
previously authenticate to the server to be able to access the data.
17
Page 17 / 95
Each message starts with a message header consisting of its length and an opcode
corresponding to a specific message type. The message is comprised of an identification
number, a channel name, and the payload.
The primary advantage of the hpfeeds protocol is its bus-like structure that makes it very
easy to connect new honeypots and receivers to the architecture. Integration only requires to
register the honeypot to the appropriate channel. Thus, there is no need to negotiate the data
exchange with all or selected sites that receive the data. In addition, a site willing to receive
data has only to register and subscribe to the channel.
4.2. Textual Data Formats
Many data exchange formats represent data in textual form. Therefore, the messages can be
processed and displayed with all programs that are able to process textual data whereas no
knowledge of the structure is required. This is an advantage compared to the previously
introduced binary data formats. The structure of the data is given by metadata such as XMLtags that are embedded in the message. Because the structuring data is part of the massage
the structure can be understood without an external specification.
XML as well as JSON introduce schemas to validate its validity. A schema is an external
document that provides a formal specification of the structure of all related documents. The
schema specifies the structure as well as the data type of each data entity. Thus, the validity
can be verified by testing if the schema meets the specification of the schema.
Because most common data exchange formats are based on either XML or JSON we here
divide all formats into these categories. XML (Extensible Markup Language) divides
characters into “markup” which structures the message and content. For example, the line
<IncidentID name="csirt.example.com">908711</IncidentID>
of an example IODEF message is comprised of the markup construct “IncidentID” and
“name” and the content “csirt.example.com” and “908711”. A formal specification of the
structure is given by a Document Type Definition (DTD) or an XML schema.
JSON (JavaScript Object Notation) is derived from the representation of simple data
structures and associative arrays in the scripting language JavaScript. In short a JSON
message consists of key and value pairs such as “firstName”: “John” whereas the first
part is the name of the data field and the second parts its value. The structure of a message
is defined by a JSON schema.
4.2.1. XML
IDMEF
The Intrusion Detection Message Exchange Format (IDMEF) as specified in [RFC 4765] is a
very versatile data format especially devoted to exchange and transmit data produced by
Intrusion Detection Systems such as Snort. The primary use case is to transmit alerts from a
sensor to a central management system on which the messages are further processed and
analysed. For example, the Prelude SIEM uses IDMEF to enable a distributed network of
sensors in that all report to one or more managers which can be hierarchically organised.
Sensors are IDSs including Snort or other components that enable to aggregate and
correlate multiple alerts. For example, this can be used to detect coordinated port-scans that
originate from more than one source.
While IDMEF messages are in principle human readable they are because of their
complexity better suited to be processed by programs. This includes, for example, the import
of data into a database where the data could be displayed by a web-application. In addition,
18
Page 18 / 95
the format is ideal to exchange alert between CSIRTs and other security-aware teams such
as ISPs.
In short, IDMEF contains the following parts:
• Identification and name of the analyser, e.g. Snort NIDS
• Time of detection and time of message generation
• Information about the source and target of an attack. This includes IP addresses, DNS
names, process IDs, and file names.
• A classification of the alert. This comprises the signature that triggered the alert.
• An assessment of the severity of the threat
• Additional data, e.g. logs or other data related to the attack
• Information about correlated alerts
IODEF
The Incident Object Description Exchange Format (IODEF) is an adoption of IDMEF that is
devoted to the exchange of computer security incidents among CSIRTs. The most important
building blocks of an IODEF message are:
• The temporal extent of the incident: This includes the time the incident has been
detected
• An assessment of the incident: A characterization of the impact of the incident.
• Method: A description of the method the attacker has been used, for example, to attack
the system under analysis.
• Contact Information: This is, for example, the postal address and telephone number of
the reporting CSIRT.
• The data related to the source and target: This contains the data concerning all sources
and targets related to the incident. For example, relevant data are the IP addresses or
DNS information of the attacking and targeted system.
• Other data related to the attack: Usually, the reporting site adds data serving as
evidence to the report. This includes log-excerpts, NetFlow data, and other information that
are related to the incident. Additionally, a complete IDMEF reports can be included.
As previously mentioned, IODEF is devoted to the exchange of security incidents by CSIRTs.
This is considered by some data entities that are specific for the requirements of this user
group. First, the incident data can include an expectation that conveys to the recipient of the
IODEF document the actions the sender is requesting. For example, this can be a request to
block a host or to prevent any further abuse. To respect privacy concerns, the disclosure of
information can be controlled by the attribute “restriction”.
TAXII and STIX
These two formats are part of a very comprehensive and versatile framework that have been
proposed by the MITRE Corporation. In short TAXII (Trusted Automated eXchange of
Indicator Information) defines a set of protocols and services to exchange cyber threat
information. The language that provides a representation of these informations is given by
19
Page 19 / 95
STIX (Structured Threat Information Expression). While IODEF focuses on the exchange of
incidents, the TAXII/STIX framework has a broader view on security incidents. A threat is
modelled by STIX comprising the following entities:
• Indicators and Observables: A specific attack typically involves pattern that allow to
characterise it. These pattern are, for example artefacts and/or behaviours of interest
within a cyber security context and are specified in STIX by Observables acting as
Indicators for an attack.
• Incidents: These are successful attacks detailing the information about the source and
target. The related Indicators and Observables of the threat give information how that
attack could be detected.
• Exploit Targets: Vulnerabilities or weaknesses that enable the attacker to successfully
attack a system.
• Tactics Techniques and Procedures (TTP): These give an overview on the overall aim of
the attack. For example, this can be to use malware to steal credentials.
• Threat Actors: A characterisation of the identity, suspected motivation, and suspected
intended effects of the attackers.
• Campaigns: Usually, attackers do not attack a single target. Instead, they target a specific
community or a set of computers or applications. This can, for example, be a set of SSH
servers that share a specific user group such as the high energy physicist.
The TAXII framework is used to share the threat information that is specified by STIX. The
framework support multiple organisational models to exchange this data. These are:
• Source-Subscriber: There is a central instance that provides the data to all consumers.
• Hub-and-Spoke: There is a central instance that provides the data to all consumers.
However, the consumers can send data to the central instance that retransmits the data.
• Peer-to-peer: There is central instance. Instead, the consumers exchange data in
arbitrarily connected networks.
Overall, the strength of TAXII and STIX is the modelling of complex attack behaviour
consisting of multiple related steps. It can be expected that this framework is well-suited for
modelling threats concerning botnets. Under this aspect, this framework is advantageous to
IODEF that servers solely as a data exchange format.
4.2.2. JSON
X-ARF
X-ARF is a light-weight but structured format for the exchange of data related to computer
security incidents. In contrast to other formats like IODEF the format is kept as simple as
possible. Thus, the aim of X-ARF is to introduce a light-weight and structured format which
focuses on the most relevant information and can easily be used and extended. The format is
not limited to incident data and can additionally be used to exchange malware, honeypot, or
IDS data. An X-ARF message contains human as well as machine readable containers.
Therefore, the same message can be used to inform the administrator of an abusive system
about the incident and it can automatically be processed by an incident management system
without changes. Currently, the format is supported by a growing list of CERTs and a broad
acceptance in the academic community can be expected.
20
Page 20 / 95
X-ARF documents are structured in multiple parts denoted as container. Each container is
structurally independent of the other and may contain completely different content. However,
the idea is to combine human readable and machine readable parts which contain the same
or at least similar information. Therefore, an X-ARF document simultaneously addresses
humans, for example, system administrators and allows an automated processing. All
specified use-cases consist of three containers, although this number may vary for future
specifications. Since X-ARF documents are transferred by e-mail, each container has a
specific MIME-type (Content-Type) and a specification of the character encoding (charset)
which is usually UTF-8. Currently, three different use cases exist for brute-force, malware and
phishing attacks that share the following containers:
• The first container is human readable and can contain arbitrary text. Its MIME type is
typically “text/plain” encoded in an UTF-8 charset.
• The second container contains data that uses the YAML markup language for structuring
the content. A JSON schema exists defining the structure and syntax of the data. This
allows to test if an X-ARF document is well-structured and valid in respect of its
specification.
• The third container is intended for transferring various additional data regarding to the
abuse type that depends on the previous specification in the second container. It can
include log-data as a kind of evidence for the abuse handler or it may be used for
malicious files that are, for example, captured by a honeypot.
The first container is intended to contain a summary of the message in textual form. The
second part contains the details of the attack data. It is structured and can be automatically
processed by an incident handling system. The fields contained in the second container
depend on the abuse type of the X-ARF document. However, all abuse types share a
common set of fields that, for example, contain data about the abusive system.
The strength of X-ARF are its simplicity and versatility. The documents contain a machine as
well as human readable part to support multiple groups of recipients. In addition, the format is
kept as simple as necessary to ease its application and assesses a lot of different use cases.
Proprietary formats based on JSON
As previously mentioned, JSON can be used to structure a message and to specify the data
types. In the ACDC community data formats exist to submit data, which, for example,
includes data gathered by honeypots. Other schema address information about hosts serving
malware URLs, location of C&C server controlling a botnet., passive DNS information,
Spambots, and IDS alerts.
The advantage of JSON is the easy and efficient definition of ad hoc or highly specialised
data formats. JSON enables the quick design of a specialised data format that is, for
example, applicable to submit data to a central repository. This is especially important, if an
appropriate data format for a specific data set such as the data mentioned above is missing.
5. Evaluation of the Data Formats
The aim of this document is to identify applicable data formats that can be used in the
context of the ACDC project. Furthermore, the document should propose extensions or
improvements if the available data format does not satisfy all requirements. These
requirements have been proposed in the second section. To identify applicable data formats
a survey has been initiated to gather information about which formats are already deployed
by the ACDC partners and what use cases are addressed. In this section these data formats
are evaluated in respect to criteria that are derived from the requirements in Section 3.
21
Page 21 / 95
A survey was conducted to collect the currently used data formats and to gather information
about the usage of these formats. This survey includes a questionnaire that is partitioned into
three blocks (for the complete questionnaire we refer to the Appendix). After the name of
data format, the second block comprises questions concerning use cases the data format is
related to. This includes the role of involved sites as well as information about incorporated
workflows. Furthermore, experiences are questioned and whether there are demands for
extensions or improvements. The third block comprises specific questions about the data
format details. This includes properties of the data format as well as any bindings to a
specific transport protocol. For example, some formats such as IODEF are designed to be
submitted by email. This is important to consider because some aspects such as the data
security are in some cases left to the protocol. For example, using S/MIME standard ensures
confidentiality, integrity and authentication of IODEF messages. A fundamental information is
whether the data is represented in binary or textual form and if there is a formal specification
of syntax and semantic. It is important to note, that such a formal specification is crucial for
an automated processing of the data. Ideally, the specification is publicly available, e.g. as an
RFC document. To ease the deployment, the availability of programs or libraries to create
and process messages in the specific data format is required. Ideally, these programs are
released under an open license such as the Gnu GPL.
5.1. Evaluation of the questionnaires
Overall, 15 responses were analysed originating from 12 different sites. Nearly all
questionnaires refer to different data formats. Only two responses refer to the same data
format (IODEF). For the complete set of anonymized responses we refer to the Appendix.
The most important result is that most of the used data formats are based on a textual
representation and are structured by XML or JSON. Among them are the publicly specified
formats X-ARF, IODEF, and STIX/TAXII that are used to exchange data with external sites.
Additionally, other proprietary formats based on JSON and XML are internally used. Other
specialised formats such as IDMEF, sFlow, and hpfeeds formats have been proposed that
are devoted to transfer data of network connections and data gathered by honeypots and
IDS such as Snort. The protocols are designed to cope with large amounts of data, for
example produced by monitoring large networks. These formats are often internally deployed
to transfer the data from a sensor such as an IDS, honeypot, or NetFlow collector to
components that aggregate and correlate the data. This process is used to combine multiple
events (e.g. IDS alerts) to the full extend of an incident.
The results of the evaluation of the questionnaires are summarised below. The
questionnaires are labelled from “A” to “Q” and grouped according to different criteria. For a
complete listing of all questionnaires we refer to Appendix. The first part summarises results
grouped by the referred data format. The next part classifies results considering a list of
major properties. In the following two parts we group the data formats according to supported
user groups and data sources. The section concludes with an enumeration of use cases that
are expected to be relevant for the project. For each use case the data formats are listed that
are applicable.
5.1.1. Overview of the received questionnaires and referred data formats
• A,B,C,D: Text based proprietary formats: Whois output, FluxDetect tool, Skanna tool,
EvidenceSeeker tool
• E,F,G: Proprietary data formats based on JSON
• H: Proprietary data formats based on XML
• I: Sflow 5.0
• J: hpfeeds
22
Page 22 / 95
• K, L: IODEF
• M: X-ARF
• N: IDMEF
• O: STIX/TAXII
• P, Q: proprietary formats
Proprietary
STIX/TAXII
IDMEF
X-ARF
IODEF
No. Questionnaires
Netflow
XML proprietary
JSON proprietary
Textual Format
0
0,5
1
1,5
2
2,5
3
3,5
4
4,5
Fig. 2: Overview of the received questionnaires
5.1.2. Overview of the technical properties of the data formats
• Textual representation: A,B,C,D,E,F,G,H,K,L,M,O,P
• Binary representation: I,J
• Machine readable: F,G,H,I,J,K,L,M,N,O
• Human readable: A,B,C,D,M,P
• Capability for Bulk-data / aggregation: G,H,I,J,M,O,P
• Formal specification of structure: E,F,G,H,I,J,K,L,M,N,O
• Public specification available: I,J,K,L,M,N,O
• Explicit support for security aspects: E,F,G,H,I,J,K,L,M,N
23
Page 23 / 95
Security
Public Spec
Formal Spec
Bulk-Data
Number of Formats
Human R
Machine R
Binary
Textual
0
2
4
6
8
10
12
14
Fig. 3: Properties of data formats
5.1.3. User groups and their requirements
• Internal usage in ACDC community: A,B,C,D,F,G,H,I,J,K,L,M,N,O,P,Q
• External data exchange with CSIRTs, ISP, Academic/Research, AV: K,L,M,N,O
– Public specification available
– Textual representation
– Security requirements met
• Law enforcement agencies: K,L,M,N,O
– Public Specification
– Textual representation
– Security requirements met
24
Page 24 / 95
Law enforcement
Data exchange
Number of Formats
Internal usage
0
2
4
6
8
10
12
14
16
Fig. 4: Number of formats supporting a specific user group
5.1.4. Overview of data formats and supported sources
• Honeypots: E,F,G,J,M,N
• NetFlow: I
• IDS: E,F,G,N
• Incident reports: K,L,M,O
• Log files: A,P
25
Page 25 / 95
Log File
Incident Report
IDS
Number of Formats
Netflow
Honeypot
0
1
2
3
4
5
6
7
Fig. 5: Number of data formats supporting a specific user group
5.1.5. Use Cases
A use case is here understood as a scenario where data is exchanged between user groups
previously introduced. This includes, for example, the submission of IDS sensor data to a
central repository. Our aim is to select typical use cases that are expected to be relevant for
the ACDC project. Although a list of general requirements is previously assessed it is
important to note, that each use case may have specific demands. For example, an
automated processing of a message requires a formal specification of structure and data
types whereas an end user report should be as simple and descriptive as possible. Thus, the
selection of a data format cannot be done without a specification of the adherent
requirements of the use case that the data format is involved into.
To retain the clarity, we use a simplified list of the requirements as described in Section 3. It is
important to note, that some requirements are taken from the specification of IODEF and
IDMEF. These requirements are explicitly satisfied by nearly all XML and JSON formats.
Furthermore, it can be expected that nearly all data formats with a formal specification
consider an unambiguous specification of encoding and data types. For the sake of clarity,
these requirements are omitted here.
Below the use case and the proposed requirements are listed. After each use case a
proposal of the applicable protocols corresponding to the labelled questionnaires are
enumerated.
• Submission of sensor data (honeypot, NetFlow, IDS): E,F,G,H,I,J,N
– Machine readable
– Formal specification to check correctness
– Data export from a sensor (e.g. honeypot)
• Data exchange to analyse
A,B,C,E,G,H,I,J,K,L,M,N,O,P
aggregate,
and
enrich
the data
(internal
usage):
– Internal description of format
26
Page 26 / 95
• Submission of incident (attack) data to central data repository: E,F,G,H,I,J,K,L,M,N
– Machine readable
– Formal specification to check correctness
– Security requirements (sensor and recipient authentication)
– Capability to anonymize or filter the data
– Support of the relevant information
– Support of access control
• Distribution of data and notification to affected/interested stakeholders: E,F,G,H,K,L,M,N
– Textual representation
– Security requirements (sender and recipient authentication)
– Public specification
– Capability to anonymize or filter the data
– Support of the relevant information
• Reporting to end users: M
– Human readable
– Public specification/description available
5.2. Summary of Results
The evaluation of the questionnaires comes to the following conclusions:
• The available data formats support the expected use cases of the ACDC project quite well.
• Shortcoming of the formats are the specification of a fine-grained access control
mechanism for specific data entities that may contain person related data. Only IODEF and
TAXII consider this. However, the granularity of the access control is very rough.
• Specialised formats are advantageous for the submission of raw data for analysis /
aggregation (sFlow, IDMEF) because they support aggregation and/or compression.
• X-ARF is the most versatile format. It is machine as well as human readable and supports
all user groups. The format is less complex than IODEF and STIX/TAXII and does not raise
the bar for usage. Therefore, it is perfectly suited to submit sensor data and to exchange
data with external sites such as ISPs, law enforcement, and end users It is important to
note, that only X-ARF provides alternative parts that address multiple user groups in
parallel.
• IODEF is specialised to exchange data among CSIRTs. It offers more features to express
expectations and restrictions on the data usage compared to other incident reporting
formats such as X-ARF. The format could be used on a bilateral basis to exchange data
with CSIRTs.
27
Page 27 / 95
• STIX/TAXII is very complex raising the bar for its usage. However, STIX provides good
means to characterise and model threats. For example, STIX can be used to model the
relationship between attackers, their methods and strategies, and observed incidents. This
can, for example, be used to analyse botnets and their characteristics.
6. Conclusion
This document aims at proposing data formats that should be adopted by the ACDC project.
To assess relevant data formats a survey has been conducted using a questionnaire
distributed to all ACDC partners. The resulting list comprises 15 data formats that can be
partitioned into 13 textual and 2 binary formats. From these formats, 10 have a formal
definition of their structure and data types which allows an automatic processing. The list of
data formats contains some well-known formats including IODEF, sFlow, and X-ARF.
In the first part of this document a list of user groups and general technical and organisational
requirements has been assessed that are relevant for data formats. In Section 5.1, these
requirements have been assigned to use cases that can be expected to be relevant for the
project. For example, such a use case is the submission of IDS sensor data to a central
repository. Furthermore, the basic requirements that are crucial for these use cases have
been stated. From this, a preliminary proposal has been given which data formats are
expected to be relevant for the individual use case. It points out that the available data
formats already cover the use cases quite well. X-ARF is the most versatile data format. Its
strength is the data exchange with external sites such as CSIRTs, ISPs, and law
enforcement. IODEF is specialised towards the data exchange with IRTs. However, the
format is more complex and requires much more effort for processing. Because of the
features especially addressed to the CSIRT community its strength is the data exchange with
selected partners that are capable of handling the format. The same is true for STIX/TAXII.
This format provides powerful means to model complex threats such as botnets. Therefore,
this format may be advantageous to store structured threat information.
Another important requirement is anonymization and access control of data. Anonymization
requires a data format in which all data types are specified. This enables to identify all data
fields containing data to be anonymized. Fortunately, IODEF as well as X-ARF and STIX
satisfy this requirement. A shortcoming of all data formats is the lack of access control.
Although IODEF and TAXII include some form of access control the granularity is very limited
and does not, for example, support different privileges for multiple user groups. However, it is
important to note that access control is often implemented by the transport protocol.
28
Page 28 / 95
7. References
D1.8.1: KU Leuven, ACDC Deliverable D1.8.1: Legal Requirements, to be published
D6.1.1: Engineering Ingegneria Informatica, ACDC Deliverable D6.1.1: User profiles and
categorization, 2013
D1.2.1: ECO, ACDC Deliverable D1.2.1: Specification of Tool Group "Centralised Data
Clearing House", to be published
RFC 5101: B. Claise, Ed., Specification of the IP Flow Information Export (IPFIX) Protocol for
the Exchange of IP Traffic Flow Information, 2008
DoW: ECO, ACDC Description of Work, 2013
RFC 4765: H. Debar, D. Curry, B. Feinstein, The Intrusion Detection Message Exchange
Format (IDMEF), 2007
29
Page 29 / 95
8. Appendix
For reference we first list the complete questionnaire that was sent to the project consortium
and afterwards the filled questionnaires with abbreviated questions.
8.1. Questionnaire
1. What is the name of the data format and which version is currently in use?
2. Specific use case:
Use cases: Please describe the specific use case or cases regarding the data exchange
format. This includes the following points:
(a)What is your role and the role of other participating sites? Why do you use the specific
format?
(b)Which workflows with respect to the import, exchange, and export of the data are
involved?
(c) Which productive software components and interfaces are used?
(d)What are your experiences? Are there any points in the format you want to improve or
are any features missing?
(e)If available, please submit any samples.
(f) Are there any licenses or patents that have to be considered concerning the application
of the data format?
3. Data format details:
Please provide us with technical details concerning the data format(s)
(a)Is there a binding to a specific Internet protocol for the transport?
(b)How is the data format structured or specified?
i. Is there a formal specification of the structure (e.g. to be machine understandable)?
ii. Is the specification publicly available? Where are they published? Are there any
standards or RFCs released providing a specification?
iii.Is it possible to extend the format?
iv.Is it possible to validate the correctness of the message syntax/semantics?
v. How is the message represented (textual, binary, or other)?
(c) Please, describe the type of data or threat for which the format is designed.
(d)Which security aspects are implemented by the data format and its related transport
protocols?
i. Confidentiality and integrity?
ii. Sender and recipient authentication?
iii.Availability?
30
Page 30 / 95
(e)Is the format adapted to the provisions of a targeted user group? If not, what are the
addressed user communities (e.g. end user, CERT, ISP).
(f) Is a specific communication infrastructure preferred?
i. Peer to peer?
ii. Centralised?
iii.Closed user group?
(g)Which software components to produce, import, export, parse, and process the data
are available? Are these publicly released? Are there any licenses or patents related to
the software that have to be considered?
31
Page 31 / 95
8.2. Questionnaire A
1. Format name and version
Our tool Evidence Seeker helps operators to extract evidences from a log file.
Input: Offline
• Plain text files, generally log files
Output:
• Plain text files. There are generated 2 files, one with contact information and the other one
with evidences
2. Use case
(a)Role and rationale
Evidence Seeker helps operators to extract evidences from a log file.
Using log files as input has the key advantage that Evidence Seeker can receive the data as
is, in the format that is generated by the application that creates the log, without any previous
manipulation.
(b)Workflows
As this tool will be part of the Centralised Data Clearing House, Evidence Seeker can be
used in any workflow where there is a need to process a log file searching for IPs suspicious
of have been compromised
(c) Software components and interfaces
The tool doesn’t use any productive software component or interface.
(d)Experiences
Plain text is a good option for evidence extraction. As logs are usually generated in plain text,
there is no need to parse the log. Other plain text advantage is that is easily understandable
and universally accepted in any operating system.
(e)Samples
(f) Licenses or patents
As the input and output are in plain text, there is not bound to any licenses or patents.
3. Format details
(a)Transport protocol
Evidence Seeker doesn’t interact with any other tool, so there is no binding to any specific
Internet protocol for the transport.
32
Page 32 / 95
(b)Structure or specification
i. Formal specification
The input and output structures follow INTECO specifications, but these specifications do not
necessarily adhere to any standard.
In order to satisfactorily process the IPs, the input file must have at the beginning of each line
the IP in numeric format.
ii. Availability of specification
The specification follows INTECO defined structure, but it does not necessary follow any
standard.
iii.Extending the format
If it is wanted to extend the format it must be taken into consideration the purpose of
Evidence Seeker.
iv.Validate syntax and semantics
The input is generally log files, so they follow a structured syntax that makes possible to
validate the input
The same happens with the output, it follows a structured syntax that makes possible to
validate it.
v. Representation
Information is saved textually in plain text.
(c) Type of data or threat
Input data are log files, designed for log event recording in a system.
Output data is designed to group IPs detected in the log file, in a structured way.
(d)Security aspects
i. Confidentiality and integrity
No, the output is in plain text without any kind of encryption or security measures
ii. Authentication
Output is generated without any consideration about the recipient.
But security restrictions can be implemented at OS level into folders where the files are
expected to be saved.
iii.Availability
Both input and output are files, that means that information is stored into file system and can
be accessed when desired.
(e)User group
Evidence Seeker is designed to facilitate the notification process obtaining evidences from a
log file, so the tool is basically thought for CERTs.
33
Page 33 / 95
(f) Communication infrastructure
Evidence Seeker currently doesn’t coordinate with any other tool or service, so there is no
specific communication infrastructure.
i. Peer-to-peer
ii. Centralised
iii.Closed user group
(g)Software components
EvidenceSeeker is offered with all the components it needs to handle information.
34
Page 34 / 95
8.3. Questionnaire B
1. Format name and version
Flux-Detect detects and monitors domains using fast-flux techniques.
INPUT:
• Plain text files, with domains lists
OUTPUT: it doesn’t generate any output; it saves information in databases.
2. Use case
(a)Role and rationale
The role of Flux-Detect is to give feedback about domains determining if they are fast-flux or
not.
(b)Workflows
Flux-Detect returns feedback about if a domain is fast-flux or not. So, both the input and
output are used or can be used in conjunction with other ACDC tools.
(c) Software components and interfaces
There are no productive software components or interfaces used.
(d)Experiences
Input and output follows our needs. Since Flux-Detect fulfils INTECO expectations, we do not
plan further improvements right now, but input and/or output can be adapted if needed in
order to integrate it with other ACDC tools.
(e)Samples
INPUT
begin
google.es
google.com
end
(f) Licenses or patents
The input is in plain text so there is no need of licenses or patents considerations. As there is
no output (information is saved in a data base) there is neither any need of licenses or
patents considerations for output.
3. Format details
(a)Transport protocol
Whois port 43 of TCP.
35
Page 35 / 95
(b)Structure or specification
i. Format specification
The input structure follows an INTECO specification, but the specification does not
necessarily adhere to any standard.
The input is a plain text file that has a list of domains to check, each of them in a different
line.
ii. Availability of specification
The input structure follows INTECO specifications, but these specifications do not
necessarily adhere to any standard.
iii.Extending the format
Flux-Detect works only with domains, so in the way the program is designed, there is no
necessity to extend the format. But it is possible to extend it if necessary.
iv.Validate syntax and semantics
As the files received must follow the structured defined, it is perfectly possible to validate the
correctness of the input files.
v. Representation
INTPUT:
•
The message is always represented textually
OUTPUT:
•
Information is saved in SQL databases
(c) Type of data or threat
INPUT: Flux-Detect receives web domains to check if they are fast-flux or not
OUTPUT: Flux-Detect determines if a domain is fast-flux or not and saves the information in
databases.
(d)Security aspects
i. Confidentiality and integrity
Flux-Detect does not implement any encryption or data security measures. But, because
there is no output as information is saved in databases, it would be possible to implement
security restrictions in the database.
ii. Authentication
There are no sender or recipient authentication implementations
iii.Availability
As information is saved in databases, it can be said that it is always available.
36
Page 36 / 95
(e)User group
There is no output format as data are saved in databases. Although the output format is not
adapted to any specific targeted group, the information may be suitable for different user
communities, for example, it may be suitable for statistical purposes, for CERTs in order to
know if a domain is fast-flux or not, etc.
(f) Communication infrastructure
The information generated by Flux-Detect, nowadays, is not publicly spread, but only
provided by a web interface to operators carrying out domain security investigation duties
i. Peer to peer
ii. Centralised
iii.Closed user group
(g)Software components
Flux-Detect is offered with all the components it needs to handle information.
37
Page 37 / 95
8.4. Questionnaire C
1. Format name and version
Skanna, checks the security level of several domains. For each domain checks several
parameters as the software installed and version.
Skanna performs the following actions:
1. Gathering of the domains to check
2. Domains information gathering and analysis
1: Domains to check gathering
The list of domains to check can be obtained in different ways:
Source
Description
Data obtained
Nic.es
Entity responsible for
the .es domains
management
.es domains list
registered since
2007
VeriSign
Obtaining of all DNS
zones of the .com,
.net and .name TLD
Used for re-scanning
The new domains
.com, .net and .name
registered daily
Domains to check
Manually
Method for
obtaining data
Download of the
published PDF file
(http), parser and
domains extraction
Reception of a file
with the domains,
one domain per line.
An operator indicates
manually the domain
to check
2: Domains information gathering and analysis
For each domain obtained in previous step, the following actions are performed:
• Obtain the index page source code of the website
• Identify the software and technologies used by the website
• Indexation of the index page source code
• Antivirus analysis of the downloaded source code, in order to identify malware or
compromise signals
To
obtain
information
about
each
domain
it
is
used
WhatWeb
(http://www.morningstarsecurity.com/research/whatweb), the information received from this
source is processed by Skanna in order to save it in databases.
The input and output are as follows:
• INPUT: domains input
• OUTPUT: there is no output. Information is saved in databases
38
Page 38 / 95
2. Use case
(a)Role and rationale
Currently Skanna doesn’t interact with other tools and/or services of ACDC. But, in order to
perform its activities, Skanna interacts with different tools and/or services external to ACDC,
and this interaction is always done requesting information to these tools/services.
(b)Workflows
The workflow is always the same; Skanna needs some information and send a request to the
tool/service needed in each moment.
(c) Software components and interfaces
Skanna interacts with the following tools/services that are not part of the ACDC project:
1. Nic.es: Skanna downloads a PDF file with the new domains
2. WhatWeb
(d)Experiences
Both input and output follow our specifications. Since Skanna fulfils INTECO expectations,
we do not plan further improvements right now, but input and/or output can be adapted if
needed in order to integrate Whois with other ACDC tools.
(e)Samples
Attached at the end of the questionnaire
(f) Licenses or patents
Skanna doesn’t use specifically any data format as it only gathers and process structured
information.
The only consideration is that WhatWeb has GPLv2 license.
3. Format details
(a)Transport protocol
The download of the PDF file from Nic.es is done by HTTP.
The interactions with the other tools/services are done locally, thus there is no information
transmission on the internet.
(b)Structure or specification
i. Format specification
Input data is obtained from different sources and the format specification is defined by each
source.
There is no output, as data are saved in different databases.
ii. Availability of specification
Input data from VeriSign is a structured file with a domain list, one per line.
39
Page 39 / 95
The information gathered from the different sources is structured. The output from WhatWeb
is in XML and follows XML standards.
There is no output, but only information saved in databases.
iii.Extending the format
It would be possible to gather information from other sources or use other formats but it
would be necessary to adapt Skanna in order to make it able to perform those new actions.
iv.Validate syntax and semantics
As all the data received are in a structured format, it would be possible to check the different
inputs using regular expressions, but nowadays there is no specific mechanism to implement
this action.
The output received from WhatWeb is in XML so it would be especially easy to validate the
correctness of the information.
v. Representation
The information is presented textually, but part of the data is saved in a DataBase following
SQL specifications.
(c) Type of data or threat
Skanna is designed to get a map about the security level of the domains inspected.
(d)Security aspects
i. Confidentiality and integrity
Skanna does not perform any confidentially or integrity checks.
There is no output as information is saved in databases, but it would be possible to
implement security restrictions in the database.
ii. Authentication
Skanna does not perform any recipient authentication
iii.Availability
The information is saved in database, so it is available when needed.
(e)User group
Skanna is a tool that aggregates relevant information about domains (technologies inventory
used by the domains, domain index and domain index analysis by an AV). This information is
provided by a web interface to operators carrying out domain security investigation duties,
allowing operator to check, search or exploit the information.
(f) Communication infrastructure
The information gathered by Skanna, nowadays, is not publicly spread, but only provided by
a web interface to operators carrying out domain security investigation duties
40
Page 40 / 95
i. Peer to peer
ii. Centralised
iii.Closed user group
(g)Software components
It is very important to note that Skanna interacts with other external tools/services needed for
a proper performance of Skanna.
Nic.es
Example
of
the
websites
http://www.dominios.es/dominios/sites/default/files/files/Altas%20abril
%202013%20%28espanol%29.pdf
for
April:
VeriSign
Domain 1
Domain 2
…
WhatWeb XML example
<?xml version="1.0"?><?xml-stylesheet type="text/xml" href="whatweb.xsl"?>
<log>
<target>
<uri>http://www.osi.es</uri>
<http-status>200</http-status>
<plugin>
<name>HTTPServer</name>
<string>Apache</string>
</plugin>
<plugin>
<name>Google-Analytics</name>
<account>UA-17786431-4</account>
</plugin>
<plugin>
<name>Apache</name>
</plugin>
<plugin>
<name>IP</name>
<string>195.235.9.101</string>
</plugin>
<plugin>
<name>JQuery</name>
</plugin>
<plugin>
<name>HTTP-Headers</name>
<string>cache-control: store, no-cache, must-revalidate, post-check=0,
pre-check=0,connection: close,content-length: 64612,content-type: text/html;
charset=utf-8,date: Tue, 14 Feb 2012 09:02:53 GMT,expires: Sun, 19 Nov 1978
05:00:00 GMT,last-modified: Tue, 14 Feb 2012 09:02:56 GMT,server: Apache,setcookie: SESS66c3c803e511690dab0e8d70f3f0cf31=oqdkjeqlv4lun5ntlprk0ut6b0;
expires=Thu, 08-Mar-2012 12:36:13 GMT; path=/; domain=.osi.es,vary: AcceptEncoding</string>
</plugin>
<plugin>
<name>Drupal</name>
</plugin>
<plugin>
41
Page 41 / 95
<name>MD5</name>
<string>b2c6f30f1355d0482e04ad869a7bd68b</string>
</plugin>
<plugin>
<name>Cookies</name>
<string>SESS66c3c803e511690dab0e8d70f3f0cf31</string>
</plugin>
<plugin>
<name>Title</name>
<string>Oficina de Seguridad del Internauta</string>
</plugin>
<plugin>
<name>Country</name>
<string>SPAIN</string>
<module>ES</module>
</plugin>
</target>
</log>
42
Page 42 / 95
8.5. Questionnaire D
1. Format name and version
Whois automates relevant IPs lookup. This service provides whois information in an efficient
and easily parseable manner.
Input:
• A single IP
• Plain text files
Output:
• Text
2. Use case
(a)Role and rationale
The role of Whois is to provide whois information. Thus, it works as a service for other ACDC
tools that need that information.
Whois output information can easily be read by humans or processed by a machine.
(b)Workflows
Whois runs in service mode (receives a request and returns an answer).
(c) Software components and interfaces
Whois queries different RIR services in order to obtain and/or update information.
It also uses a free database of Maxmind in order to obtain the country an IP belongs to.
(d)Experiences
Both input and output follows our specifications. Since Whois fulfils INTECO expectations,
we do not plan further improvements right now, but input and/or output can be adapted if
needed in order to integrate Whois with other ACDC tools.
(e)Samples
INPUT
• Plain text file with IPs
begin
verbose
8.8.8.8
193.245.3.4
end
OUTPUT
• IPS text response
43
Page 43 / 95
15169 | 8.8.8.8 | 8.8.8.0/24 | US | Arin | cfg:[email protected] cpg:[email protected] r:[email protected] r:[email protected] r:[email protected] r:[email protected] | GOOGLE - Google Inc.
6848 | 193.245.3.4 | 193.244.0.0/15 | BE | Ripe | cg:[email protected]
cg:[email protected] r:[email protected] | TELENET-AS Telenet N.V.
(f) Licenses or patents
Both input and output are textual, and it is also possible to receive the input in a file in plain
text, so there is no need of license or patents considerations.
3. Format details
(a)Transport protocol
It receives input files through Whois port (TCP 43)
(b)Structure or specification
INPUT
• IPs file
The file must be created according to the following format:
First line: begin
Second line: parameters
IP addresses, one per line
Last line: end
OUTPUT
• IPS query.
The output in response to an IP query is as follows:
• AS (Autonomous System) Number
• IP address
• CIDR (Classless Inter-Domain Routing)
• Country code
• RIR (Regional Internet Registry) the IP belongs to
• IP contacts, with different TAGS
• AS (Autonomous System) Name
i. Format specification
The input and output structures follow INTECO specifications, but the specifications do not
necessarily adhere to any standard.
44
Page 44 / 95
Our own specification is the one showed before, the input is a single IP or a file in plain text,
and the output are several fields separated by the character |
ii. Availability of specification
The input and output structures follow INTECO specifications, but the specifications do not
necessarily adhere to any standard.
iii.Extending the format
If it is wanted to extend the format, it must be taken into consideration the purpose of Whois
and that the only data that it receives are IPs.
iv.Validate syntax and semantics
As the input and output follow a strict syntax, yes, it would be possible to validate the
correctness of the message
v. Representation
The output is represented textually.
(c) Type of data or threat
Output is designed to provide contact information about an IP. The output has several fields,
each of them with different information about the IP, but arranged following a fixed structured
showed before, in order to be easily understandable.
This information can be obtained either by an operator through command line interface or by
another program.
(d)Security aspects
i. Confidentiality and integrity
Neither the data format nor its related transport protocols support any security measures
ii. Authentication
Whois may check the origin of the query (will check the IP) and return more or less
information depending on the questioner
iii.Availability
There is no special protection for availability
(e)User group
Whois is suitable for any user/program, but basically focused on CERTs, that need to know
the contact data for IPs.
(f) Communication infrastructure
Whois currently doesn’t coordinate with any other tool or service, so there is no specific
communication infrastructure.
45
Page 45 / 95
i. Peer to peer
ii. Centralised
iii.Closed user group
(g)Software components
Whois is offered with all the components it needs to handle information.
46
Page 46 / 95
8.6. Questionnaire E
1. Format name and version
Suricata engine is being used as a NIDS engine on a wireless AP, which is used as a
gateway for mobile devices. The NIDS engine allows us to monitor and analyse network
traffic of mobile devices running over wireless AP. The traffic can be captured in PCAP format
and, moreover, off-line (almost realtime) analysis of PCAP files is possible. Additionally,
logging to database with possibility of e-mail notifications is also possible. There are multiple
possible log outputs (configurable):
• Line based alerts log (fast.log)
• Log output for use with Barnyard (unified.log)
• Alert output for use with Barnyard (unified.alert)
• Packet log (pcap-log)
• Files log (json format)
For us most important is files-json.log which holds data for every single file that crossed
your http pipe. Using additional fuse file-system library (e.g. ClamAV1) we can integrate other
tools for further analysis of the traffic captured.
2. Use case
(a)Role and rationale
We use the described formats in order to easily analyse the output from the NIDS and import
it into HBase database running on Hadoop. Additional analytics can be done over HBase
database (for further big-data analytics).
(b)Workflows
The output from NIDS can be transformed from CSV or JSON string formats practically into
any format data (e.g. TAB delimited format) that is needed to be transported to other module
in the workflow. We are using Flume to scan /tmp/logs directory for parsed files-json.log
files and stores them into HBase database for further analysis.
(c) Software components and interfaces
Suricata engine, Flume as a transportation level of captured data in HBase; data in HBase is
ready for further analysis. Google Cloud Messaging is used to push messages towards
mobile clients.
(d)Experiences
Not yet clear about missing features. Mow, we are able to detect some anomalies (e.g.
possible scans from mobile devices, detection of downloading of malware software – with the
use of third-party tool for analysing the malware content of downloaded packages).
(e)Samples
An example of package detection while downloading specific package from the Android
marketplace.
1 http://www.clamav.net/lang/en/
47
Page 47 / 95
{
"timestamp": "04\/25\/2013-10:01:22.552241",
"ipver": 4,
"srcip": "173.194.70.100",
"dstip": "172.16.118.69",
"protocol": 6,
"sp": 80,
"dp": 54356,
"http_uri": "\/market\/download\/Download?
packageName=com.overlook.android.fing&versionCode=210&token=AOTCm0QMRhNQICVmjtrRg-uK3lCqsg4kqRcfv4Mp40sMxtyZ4B9I0X1_ksrJbGpNyz3PIwGJWPUDcbaTSc6JUz28gTuDkp5srwtfV5vf0&d
ownloadId=1108987573357907795",
"http_host": "android.clients.google.com",
"http_referer": "<unknown>",
"http_user_agent": "AndroidDownloadManager\/4.2.2 (Linux; U; Android 4.2.2;
Galaxy Nexus Build\/JDQ39)",
"filename": "\/market\/download\/Download",
"magic": "unknown",
"state": "CLOSED",
"stored": false,
"size": 572
}
An example of PDF file detection while downloading the file using mobile device.
{
"id": 8,
"timestamp": "05\/08\/2013-13:50:21.732132",
"ipver": 4,
"srcip": "173.1.226.155",
"dstip": "192.168.14.201",
"protocol": 6,
"sp": 80,
"dp": 47101,
"http_uri": "\/pdfs\/PrimoPDF_V5_User_Guide.pdf",
"http_host": "www.primopdf.com",
"http_referer": "http:\/\/www.google.si\/search?
q=pdf+manual&ei=MS6KUamIM8m1PM6zgMgF&start=10&sa=N&biw=360&bih=567",
"http_user_agent": "Mozilla\/5.0 (Linux; U; Android 4.2.2; en-us; Galaxy
Nexus Build\/JDQ39) AppleWebKit\/534.30 (KHTML, like Gecko) Version\/4.0
Mobile Safari\/534.30",
"filename": "\/pdfs\/PrimoPDF_V5_User_Guide.pdf",
"magic": "PDF document, version 1.4",
"state": "UNKNOWN",
"stored": true,
"size": 25736
}
(f) Licenses or patents
No. Format is JSON and a result of an open source engine.
3. Format details
(a)Transport protocol
No. However, HTTP is usually used with JSON.
48
Page 48 / 95
(b)Structure or specification
i. Format specification
The format is in JSON and there is no formal specification of the data format. However, we
are providing informal data format here.
{
"timestamp": <time stamp>,
"ipver": <ip version>,
"srcip": <source IP>,
"dstip": destination IP>,
"protocol": <protocol id - 6-TCP>1,
"sp": <source port>,
"dp": <destination port>,
"http_uri": <uri part after http_host>,
"http_host": <host>,
"http_referer": <link from which source accessed the destination>,
"magic": <file command's magic pattern file>2,
"state": "CLOSED",
"md5": <md5 hash of the file>,
"stored": <was the file stored on file system>,
"size": <file size>
}
ii. Availability of specification
There is no formal specification of the output format. However, the input (
iii.Extending the format
The format can be extended using plugins or addons after the log has been created.
iv.Validate syntax and semantics
It can be validated with a simple JSON validation program or script.
v. Representation
It is represented as text.
(c) Type of data or threat
The format is designed to describe every single file that crosses configured HTTP pipe and is
(can be) captured by Suricata’s engine.
(d)Security aspects
i. Confidentiality and integrity
It is core data format and is not being exchanged with external components (yet). It is used
by the component of the framework for mobile devices security.
ii. Authentication
None security aspects are implemented by the data format in this aspect.
1 http://en.wikipedia.org/wiki/List_of_IP_protocol_numbers
2 Same output as “file” or “magic” command
49
Page 49 / 95
iii.Availability
None security aspects are implemented by the data format in this aspect.
(e)User group
The format is not adapted to the provisions of a target user groups. The format presents the
foundation of the information produced by additional analysis tools taking into account data
captured within files-json. The results of the analysis are sent to CERTs and possibly ISP for
further analysis. We are not using any other format for exchanging information with external
entities.
(f) Communication infrastructure
i. Peer to peer
ii. Centralised
iii.Closed user group
Preferred communication infrastructure is centralized since we need central endpoint to
aggregate information from the files-json. However, the architecture of the system under the
aggregation end-point can be designed to be highly available and distributed.
(g)Software components
There is plethora of open-source tools available for processing the JSON data format (python
libraries, libraries for java). All are publicly available and easily extensible. There are no
licences or patents related to the software. The use of custom created script with MySQL or
PostgreSQL import (bulk or continuous) or importing it directly to MongoDB (native import of
JSON files) are already available on the web page of Suricata 1. As already described,
Apache Flume2 framework can be used to import output (files-json) into big-data framework
for further analitics.
1 https://redmine.openinfosecfoundation.org/projects/suricata/wiki/What_to_do_with_filesjsonlog_output
2 http://flume.apache.org/FlumeDeveloperGuide.html
50
Page 50 / 95
8.7. Questionnaire F
1. Format name and version
JSON
2. Use case
JSON is a general-purpose data format to exchange information between two entities.
(a)Role and rationale
We run a set of different Honeypots and additional passive sensors. The gathered
information of all sensors has to be correlated into a single report representing individual
incidents. In order to not lose any information, the utilized data format needs to be able to
hold all information generated by the used set of tools. Since we cannot forecast the future
and anticipate any future information that might be generated by updated or new tools, the
data format has to be flexible to hold arbitrary future data as well. As a result, we opted for
JSON, which is fully flexible, human and machine-readable, produces only little overhead
and is out-of-the box supported by major programming languages.
(b)Workflows
A set of different tools generates individual JSON reports that are sent to a correlation server.
This server correlates all reports belonging to the same incident and forwards the information
to subscribed clients. One of these clients stores generated reports in a NoSQL database
(MongoDB) that also handles JSON natively.
(c) Software components and interfaces
We use internal implementations to generate JSON reports from particular honeypots and
passive sensors. These include p0f, snort, dionaea, glaspot and kippo. Sensors developed
by us support JSON natively. No additional software is required for parsing JSON messages
in python, for java we use Jackson.
(d)Experiences
Like any other text-based reporting format, JSON is rather inefficient for transmitting binary
data since it has to be encoded (e.g. base64).
(e)Samples
Attached below.
(f) Licenses or patents
No
3. Format details
(a)Transport protocol
No. JSON can be transmitted by using arbitrary transport protocols.
(b)Structure or specification
i. Format specification
Yes. http://tools.ietf.org/html/rfc4627
51
Page 51 / 95
ii. Availability of specification
Yes. http://tools.ietf.org/html/rfc4627
iii.Extending the format
According to the RFC “A JSON parser MAY accept non-JSON forms or extensions.”. Anyway,
this would rather be an exception since it is generally not necessary to extend the format
itself.
iv.Validate syntax and semantics
Yes. This is done by default libraries of many programming languages and can be done by
various other tools.
v. Representation
Pure textual.
(c) Type of data or threat
JSON is not threat-bound. It is used for arbitrary data and was originally designed to
represent JavaScript objects.
(d)Security aspects
i. Confidentiality and integrity
None. Security aspects have to be implemented by underlying transport protocols, like SSL.
ii. Authentication
None. Security aspects have to be implemented by underlying transport protocols, like SSL.
iii.Availability
None. Security aspects have to be implemented by underlying transport protocols, like SSL.
(e)User group
The format is general-purpose.
(f) Communication infrastructure
i. Peer to peer
ii. Centralised
iii.Closed user group
Communication infrastructure solely depends on the underlying transport protocol, which is
completely independent from JSON.
(g)Software components
An extensive list of software components can be found here: http://www.json.org/index.html
(you need to scroll down a little bit)
52
Page 52 / 95
Sample JSON message
{
"endtime": {
"$date": 1368970600734
},
"whois": {
"cc": "RU",
"owner": "MORDOVIA-AS OJSC Rostelecom",
"BGP_prefix": "87.119.224.0/19",
"asn": 34449
},
"geoip": {
"city": "Saransk",
"region_name": "Mordovia",
"region": "46",
"area_code": 0,
"time_zone": "Europe/Samara",
"longitude": 45.18330001831055,
"metro_code": 0,
"country_code3": "RUS",
"latitude": 54.18330001831055,
"postal_code": null,
"dma_code": 0,
"country_code": "RU",
"country_name": "Russian Federation"
},
"remotehost": "XX.XX.XX.XX",
"connections": [
{
"connection_type": "accept",
"remoteport": 2001,
"p0f_profile": {
"uptime": "-1",
"dist": "15",
"fw": "0",
"tos": "",
"detail": "2000 SP4, XP SP1+",
"link": "IPv6/IPIP",
"nat": "0",
"genre": "Windows"
},
"protocol": "smbd",
"localport": 445,
"starttime": {
"$date": 1368970592858
},
"endtime": {
"$date": 1368970593438
},
"transport": "tcp"
},
{
"remoteport": 2013,
"endtime": {
"$date": 1368970595108
},
"localport": 139,
"starttime": {
"$date": 1368970594042
}
},
{
"connection_type": "accept",
"remoteport": 2004,
"p0f_profile": {
"uptime": "-1",
53
Page 53 / 95
"dist": "15",
"fw": "0",
"tos": "",
"detail": "2000 SP4, XP SP1+",
"link": "IPv6/IPIP",
"nat": "0",
"genre": "Windows"
},
"protocol": "smbd",
"localport": 445,
"downloads": [
{
"peid": {},
"virustotal": {
"date": 1368904680,
"report": {
"Microsoft": "Worm:Win32/Gnoewin.A",
"Norman": "Inject.AQTC",
"Panda": "Suspicious file",
"ESET-NOD32": "a variant of Win32/Injector.AFKU",
"VBA32": "Worm.VBNA"
},
"ratio": 5
},
"url": "https://hotfile.com/dl/223458246/4bd6f53/g1.exe",
"md5hash": "0a0375431f8d125bfc12950abd98876e",
"peXaminer": {
"File Statistics": {
"Attributes": {
"created": "Sun May 19 13:42:51 2013",
"file_name":
"/data/binaries/0a0375431f8d125bfc12950abd98876e",
"last_accessed": "Sun May 19 13:42:51 2013",
"last_modified": "Sun May 19 13:42:51 2013",
"entropy": 7.318794553483753,
"file_size": 115633
},
"Hashes": {
"sha256":
"e69a9c7e442adb837f7af1d3a935965623b9d4354d68b66b21b28ae75b430847",
"sha512":
"4679cd287cd2ebd62d56e510ac20d88ae04b02966dfad21fcb0090b8421ddc075fac3ca769296
b70bd1fd4f5569a61be5532116ded0fc196508d4e305a58f258",
"md5": "0a0375431f8d125bfc12950abd98876e",
"sha1":
"3d7614ca28f924e459c3e86c0b661021f01c54b1"
}
},
"PE Characteristics": {
"Optional Header": {
"SectionAlignment": 4096,
"SizeOfCode": 35328,
"Magic": "32bit",
"SizeOfUninitializedData": 0,
"MinorSubsystemVersion": 1,
"MajorLinkerVersion": 10,
"ImageBase": 4194304,
"SizeOfInitializedData": 18944,
"SizeOfImage": 77824,
"NumberOfRvaAndSizes": 16,
"FileAlignment": 512,
"MajorSubsystemVersion": 5,
"CheckSum": {
"given": 116746,
"true": 120800
},
"Subsystem": "GUI",
"MinorLinkerVersion": 0,
54
Page 54 / 95
"AddressOfEntryPoint": 6351,
"SizeOfHeaders": 1024
},
"File Header": {
"TimeDateStamp": {
"UTC": "Sat May 18 01:25:29 2013",
"numerical": 1368840329
},
"Machine": "i386",
"Characteristics": [
"Executable Image",
"32bit"
],
"NumberOfSymbols": 0,
"NumberOfSections": 5,
"SizeOfOptionalHeader": 224
},
"DOS Header": {
"e_lfanew": 224
},
"Sections": [
{
"Name": " .text",
"Characteristics": [
"execute",
"read"
],
"SizeOfRawData": 35328,
"Entropy": 6.5357099216549095,
"VirtualSize": 35192,
"VirtualAddress": 4096,
"PhysicalAddress": 1024,
"md5": "729dfe04aad1c60369dec9455decd4ed"
},
{
"Name": " .rdata",
"Characteristics": [
"read"
],
"SizeOfRawData": 9216,
"Entropy": 4.772166382739247,
"VirtualSize": 9128,
"VirtualAddress": 40960,
"PhysicalAddress": 36352,
"md5": "805bc471f9d81754b3780a657d5c2f14"
},
{
"Name": " .data",
"Characteristics": [
"read"
],
"SizeOfRawData": 4096,
"Entropy": 2.1265588781733644,
"VirtualSize": 15680,
"VirtualAddress": 53248,
"PhysicalAddress": 45568,
"md5": "34ba24583e66905e5c218214d52df071"
},
{
"Name": " .rsrc",
"Characteristics": [
"read"
],
"SizeOfRawData": 2048,
"Entropy": 5.129072542887932,
"VirtualSize": 1764,
"VirtualAddress": 69632,
"PhysicalAddress": 49664,
55
Page 55 / 95
"md5": "2d470a068fec565e16520f1ffb5f13a4"
},
{
"Name": " .reloc",
"Characteristics": [
"read"
],
"SizeOfRawData": 3584,
"Entropy": 4.933846775740402,
"VirtualSize": 3366,
"VirtualAddress": 73728,
"PhysicalAddress": 51712,
"md5": "77c55c138cb3daed098db14f48b15e49"
}
],
"Data Directories": {
"Imports": {
"Descriptors": [
{
"KERNEL32_dll": [
"LockResource",
"LoadResource",
"FindResourceA",
"GetProcAddress",
"GetModuleHandleA",
"Sleep",
"GetCommandLineA",
"HeapSetInformation",
"HeapAlloc",
"SetUnhandledExceptionFilter",
"GetModuleHandleW",
"ExitProcess",
"DecodePointer",
"WriteFile",
"GetStdHandle",
"GetModuleFileNameW",
"GetModuleFileNameA",
"FreeEnvironmentStringsW",
"WideCharToMultiByte",
"GetEnvironmentStringsW",
"SetHandleCount",
"InitializeCriticalSectionAndSpinCount",
"GetFileType",
"GetStartupInfoW",
"DeleteCriticalSection",
"EncodePointer",
"TlsAlloc",
"TlsGetValue",
"TlsSetValue",
"TlsFree",
"InterlockedIncrement",
"SetLastError",
"GetCurrentThreadId",
"GetLastError",
"InterlockedDecrement",
"HeapCreate",
"QueryPerformanceCounter",
"GetTickCount",
"GetCurrentProcessId",
"GetSystemTimeAsFileTime",
"MultiByteToWideChar",
"ReadFile",
"UnhandledExceptionFilter",
"IsDebuggerPresent",
"TerminateProcess",
"GetCurrentProcess",
"EnterCriticalSection",
56
Page 56 / 95
"LeaveCriticalSection",
"IsProcessorFeaturePresent",
"SetFilePointer",
"RtlUnwind",
"LoadLibraryW",
"HeapFree",
"GetCPInfo",
"GetACP",
"GetOEMCP",
"IsValidCodePage",
"SetStdHandle",
"GetConsoleCP",
"GetConsoleMode",
"FlushFileBuffers",
"CloseHandle",
"CreateFileW",
"HeapSize",
"HeapReAlloc",
"LCMapStringW",
"GetStringTypeW",
"WriteConsoleW",
"SetEndOfFile",
"GetProcessHeap"
]
}
],
"NumberOfImports": 70
},
"Resources": {
"number_of_resources": 3,
"total_size": 0,
"entries": [
{
"type": "",
"sub_entries": 2,
"size": 0
},
{
"type": "RT_VERSION",
"sub_entries": 1,
"size": 0
},
{
"type": "RT_MANIFEST",
"sub_entries": 1,
"size": 0
}
]
}
}
}
},
"mime": "PE32 executable for MS Windows (GUI) Intel 80386
32-bit",
"ssdeep":
"3072:gV6BJx9epPREuGO7CERO9dBZiAUW4HnnnshDHV:o6BJx9epP+71ZirxMd1"
}
],
"smb_profile": {
"smb_dcerpc_requests": [
{
"dcerpcrequest_uuid": "4b324fc8-1670-01d3-12785a47bf6ee188",
"dcerpcrequest_opnum": 31
}
],
"smb_dcerpc_binds": [
{
57
Page 57 / 95
"dcerpcbind_uuid": "b3332384-081f-0e95-2c4a302cc3080783",
"dcerpcbind_transfersyntax": "8a885d04-1ceb-11c9-9fe808002b104860"
},
{
"dcerpcbind_uuid": "a71e0ebe-6154-e021-91045ae423e682d0",
"dcerpcbind_transfersyntax": "8a885d04-1ceb-11c9-9fe808002b104860"
},
{
"dcerpcbind_uuid": "7f4fdfe9-2be7-4d6b-a5d4aa3c831503a1",
"dcerpcbind_transfersyntax": "8a885d04-1ceb-11c9-9fe808002b104860"
},
{
"dcerpcbind_uuid": "d89a50ad-b919-f35c-1c994153ad1e6075",
"dcerpcbind_transfersyntax": "8a885d04-1ceb-11c9-9fe808002b104860"
},
{
"dcerpcbind_uuid": "9f7e2197-9e40-bec9-d7eba4b0f137fe95",
"dcerpcbind_transfersyntax": "8a885d04-1ceb-11c9-9fe808002b104860"
},
{
"dcerpcbind_uuid": "8b52c8fd-cc85-3a74-8b1529e030cdac16",
"dcerpcbind_transfersyntax": "8a885d04-1ceb-11c9-9fe808002b104860"
},
{
"dcerpcbind_uuid": "9acbde5b-25e1-7283-1f10a3a292e73676",
"dcerpcbind_transfersyntax": "8a885d04-1ceb-11c9-9fe808002b104860"
},
{
"dcerpcbind_uuid": "c0cdf474-2d09-f37f-beb873350c065268",
"dcerpcbind_transfersyntax": "8a885d04-1ceb-11c9-9fe808002b104860"
},
{
"dcerpcbind_uuid": "ea256ce5-8ae1-c21b-4a17568829eec306",
"dcerpcbind_transfersyntax": "8a885d04-1ceb-11c9-9fe808002b104860"
},
{
"dcerpcbind_uuid": "7d705026-884d-af82-7b3d961deaeb179a",
"dcerpcbind_transfersyntax": "8a885d04-1ceb-11c9-9fe808002b104860"
},
{
"dcerpcbind_uuid": "4b324fc8-1670-01d3-12785a47bf6ee188",
"dcerpcbind_transfersyntax": "8a885d04-1ceb-11c9-9fe808002b104860"
}
]
},
"transport": "tcp",
58
Page 58 / 95
"starttime": {
"$date": 1368970593019
},
"download_offers": [
{
"url": "https://hotfile.com/dl/223458246/4bd6f53/g1.exe"
}
],
"endtime": {
"$date": 1368970600734
},
"emu_profile": [
{
"return": "0x7df20000",
"args": [
"urlmon"
],
"call": "LoadLibraryA"
},
{
"return": "0",
"args": [
"",
"https://hotfile.com/dl/223458246/4bd6f53/g1.exe",
"20.exe",
"0",
"0"
],
"call": "URLDownloadToFile"
},
{
"return": "32",
"args": [
"20.exe",
"895"
],
"call": "WinExec"
},
{
"return": "0",
"args": [
"-1"
],
"call": "Sleep"
}
]
}
],
"flags": [
"download",
"dlserver",
"scan_vertical"
],
"dns": "87-119-XX-XX.saransk.ru.",
"_id": {
"$oid": "5198d6db0cf2f2bc1bd5cb34"
},
"starttime": {
"$date": 1368970592858
},
"localhost": "XX.XX.XX.XX"
}
59
Page 59 / 95
8.8. Questionnaire G
1. Format name and version
JSON
2. Use case
(a)Role and rationale
We are using JSON format since it is native output from our ACDC components implemented
in Python.
Our role is sending bulk reports about malware URLs, C&C, fast-flux domains and spam
campaigns. We use it because it allows large quantities of data at once without much
overhead. When used over SSL, it is not limited by maximum allowed attachment size of
SMTP servers
(b)Workflows
We inport various data formats into our system and export data from local DB in JSON
format.
(c) Software components and interfaces
We import several data formats by our components, but the role of mediation server is to
normalize all imported data into unique format and this data is then exported to Central
Clearing House in JSON format according to our schema. Mediation server and JSON are
the only interface to CCH.
(d)Experiences
Experiences with JSON are OK.
We still have not defined how to encode binary samples to be transferred to Central Clearing
House
(e)Samples
Samples are in the attachment
(f) Licenses or patents
No
3. Format details
(a)Transport protocol
There is no binding, but our preferred transport is JSON over SSL
(b)Structure or specification
i. Format specification
JSON schema
ii. Availability of specification
No, specification is defined by us.
60
Page 60 / 95
iii.Extending the format
Yes
iv.Validate syntax and semantics
Yes
v. Representation
Textual
(c) Type of data or threat
List of fast-flux domains, list of malware URLs, list of spambots in spam campaigns, list of IP
addresses related to botnet C&C
(d)Security aspects
i. Confidentiality and integrity
Is related to SSL
ii. Authentication
Is related to SSL
iii.Availability
No
(e)User group
User of this data is ACDC project.
(f) Communication infrastructure
i. Peer to peer
No
ii. Centralised
Yes, we are sending data to CCH
iii.Closed user group
No
(g)Software components
Any language supporting JSON. There are no licences or patents
Attacker data and attacking malware(usually exploit(S))
"HoneypotAttackersData"={
"AttackerData": [
"timestamp": "2013-04-29 14:02:38",
61
Page 61 / 95
"attackerIP": "5.34.247.100",
"srcPort": "58063",
"dstPort": "80",
"protocol": "http",
"countryCode": "None" ,
"sample": ["902fe4a680a1b42cdba57c551b32c13b", ""]
"compromisedURL": ["http://Jinn-tech.com/wikka/DinosgVealpr
%3ERecommended+Resource+site%3C/a%3E", ""]
]
}
Hosts serving Malware URL, phishing or C&C
"CompromisedHostsData"={
"CompromisedHost": [
"IP": "62.73.4.10",
"domain": "heuro-vacances.fr",
"country": "FR",
"type":"malware|c&c|phishing"
"malwareData":[
{
"timestamp": "2013-04-30 07:03:42.530230",
"infectedURLs": ["heuro-vacances.fr/5nW.exe","",""]
}
]
]
}
samples
"SamplesData"={
"sample": [
"timestamp": "2013-04-29 14:02:38",
"compromisedHost":"url|attachment",
"source":"spamtrap|honeypot",
"data":{
"attackerIP": "5.34.247.100",
"protocol": "http",
"countryCodeIP": "None",
"checksum":"9e3185c2dfed567442cddf466f20f9a0"
}
]
}
Passive DNS replication(fast flux domains)
"pDNSData" = {
"domains": [
{ "domain" : {
"domain_name": "example.ru",
"time_first": "2012-01-10 16:45",
"time_last": "2012-01-10 16:45",
"ips": [["IP":["121.454.32.23", "198.193.53.141"], "timestamp":
"2012-01-10 16:45:00 UTC"],
["IP":["132.123.193.23", "198.193.46.1"], "timestamp": "201201-10 16:55:00 UTC"]]
}}
]
}
Spambots participating in detected spam campaigns
62
Page 62 / 95
"spamtrapCampaigns"={
"campaign":[{
"startTimestamp":"2012-01-10 16:45",
"endTimestamp":"2012-01-12 19:45",
"total_spams":"22",
"spamSubject":"Teik it or leave it",
"has_malware":"1",
"spambot":[
{ "ip":"127.0.0.1"
"asn":"2108"
"timestamp":"2012-01-10 16:45",
}]
}
]
}
63
Page 63 / 95
8.9. Questionnaire H
1. Format name and version
Out tool, MMT, allows monitoring and analysing network traffic and any structured data (logs,
business activity, messages...). It is composed of several modules: MMT_Core that does
data extraction (e.g., using DPI); MMT_QoS/QoE that does performance analysis; and,
MMT_Security that analyses data to detect anomalous behaviour. Functionality can be
extended via plugins.
Input (offline, i.e., reading a file, or online, listening to a network interface)
1. PCAP v2.4
2. Any structured data (by writing a plugin)
Output:
1. CSV
2. Database tables (e.g., PostgresSQL)
3. XML
4. Any format (by adapting the main)
2. Use case
(a)Role and rationale
The role of our tool is to analyse traffic data to recuperate information that can be useful for
detecting botnets and other abnormal or malicious behaviour. The tool can be installed
anywhere to analyse network interfaces to generate reports (e.g., alarms or messages) that
can be sent to any stakeholder using any means (HTTP/RESTful, emails, SQL data...). The
reports contain information that can be read by humans or processed by a machine.
(b)Workflows
The tool that can be used as part of any workflow where there is a need to analyse structured
data or communication protocol exchanges and generate reports.
(c) Software components and interfaces
Modules developed by us that use the PCAP interface for network traffic extraction.
(d)Experiences
It is possible to create new plugins to analyse new types of data or adapt the main to produce
new results with different formats. The analysis that is performed is based on a given set of
security rules. These rules need to be carefully specified to avoid detecting to many false
positives or to few true negatives.
(e)Samples
1. Detection of malicious nodes in an ad-hoc network:
Input:
64
Page 64 / 95
• TDMA, Time Division Multiple Access, protocol traces from OSI layer 1+2 generated by a
Omnet++ simulator in ASCII+HEXA format, as for instance:
TS=5: smac[0x0002]: Reception SPHY_DATA_IND(SCH)
0000 01 2001 0001
00000005 0000 00 00 0030 0E 000014F0 00000000 000007D0 000007D0 00000000
00000003 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 01 00 00 08 10 10 00 08 0A 02 00 02 000200 000100
...
Output:
<?xml version="1.0" encoding="ISO-8859-1"?>
<?xml-stylesheet type="text/xsl" href="results.xsl"?>
<results>
<detail>
<occurence>
<property_id>1</property_id >
<verdict>not_respected</verdict>
<description>
ATTACK: A node is repeatedly sending MSG_SPHY_DATA_IND messages using
incorrect slots, provoking repeated slot reallocation. Could be interpreted as
a DoS attack.
</description>
<!--description of events that triggered the rule -->
<event>
<attribute><attribute_value>- - - - -timeslot=000005</attribute_value></attribute>
<description>EVENT: MSG_SPHY_DATA_IND message received</description>
<attribute><attribute_value>- - - - - - MSG_SPHY_DATA_IND.ADDRESS_SOURCE =
10:10:00:08:0A:02:00:00</attribute_value></attribute>
<attribute><attribute_value>- - - - - - BASE.TIME_SLOT =
5</attribute_value></attribute>
<attribute><attribute_value>- - - - - - MSG_SPHY_DATA_IND.SLOT_ID =
1</attribute_value></attribute>
<attribute><attribute_value>- - - - - - MSG_SPHY_DATA_IND.SLOT_TYPE =
0</attribute_value></attribute>
<attribute><attribute_value>- - - - - - BASE.PROTO =
801</attribute_value></attribute>
</event>
<event>
<attribute><attribute_value>- - - - - timeslot=000005</attribute_value></attribute>
<description>EVENT: MSG_SPHY_DATA_IND messages must to be separated by 50
slots</description>
<attribute><attribute_value>- - - - - - MSG_SPHY_DATA_IND.ADDRESS_SOURCE =
10:10:00:08:0A:02:00:00</attribute_value></attribute>
<attribute><attribute_value>- - - - - - BASE.TIME_SLOT =
5</attribute_value></attribute>
<attribute><attribute_value>- - - - - - MSG_SPHY_DATA_IND.SLOT_ID =
30</attribute_value></attribute>
<attribute><attribute_value>- - - - - - MSG_SPHY_DATA_IND.SLOT_TYPE =
0</attribute_value></attribute>
<attribute><attribute_value>- - - - - - BASE.PROTO =
801</attribute_value></attribute>
</event>
</occurence>
...
That viewed with a browser gives something like this:
65
Page 65 / 95
(f) Licenses or patents
It is not bound to any licenses or patents.
3. Format details
(a)Transport protocol
No. The tool can detect and analyse more than 600 different protocols (that includes all the
most common internet protocols and web applications), and more can be added if necessary.
(b)Structure or specification
i. Format specification
Input and output data is formally specified and can be machine processed.
ii. Availability of specification
Input data is specified by IETF (in the case of Internet protocols) or could be specific to
certain applications/services/systems (in the case of, e.g., Business Activity Monitoring).
Output data is formally specified but is defined as needed and does not necessarily follow
any standards.
iii.Extending the format
Both input and output can be extended to include new formats.
iv.Validate syntax and semantics
In most cases, yes, tools exist that can validate the correctness of input and output.
v. Representation
As preferred: textual, binary, XML, SQL...
(c) Type of data or threat
Output data is designed to detect any abnormal behaviour. For this, security properties or
rules need to be defined that describe the sequence of events that can be considered a
66
Page 66 / 95
vulnerability or a threat. The tool will use these rules to detect occurrences of these
sequences in the input and produce the results as output.
The security properties (that can be considered as internal data used by the tool) are written
using a proprietary XML format. They can be specified by us or by others but require very
good knowledge of the input that will be analysed and what can be considered correct or
incorrect behaviour.
(d)Security aspects
i. Confidentiality and integrity
Output data can be encrypted.
ii. Authentication
Using public key encryption.
iii.Availability
Depends on the communication channel used.
(e)User group
Yes, to any user that needs to analyse communication traffic.
(f) Communication infrastructure
No special preferences, supports all communication infrastructures.
i. Peer to peer
Ok
ii. Centralised
Ok
iii.Closed user group
Ok
(g)Software components
A version of the MMT_Core module will be made available as freeware. This module
captures and extracts the data needed from the input.
A version of the MMT_Security module will be available as Open Source. This module
analysed the date extracted by MMT_Core and produces the output. Depending on the
format of this output, other freely available tools probably exist that can be used to visualize
or process it
The MMT_QoS/QoE module will be available only through licensing or special agreements.
Commercial use of any of the modules is subject to licensing or special agreements.
67
Page 67 / 95
8.10. Questionnaire I
1. Format name and version
Sflow 5.0 – will be replaced by IPFix later this year
2. Use case
Exporting of Sflow samples.
(a)Role and rationale
Because its the only format our hardware supports
(b)Workflows
Receivers of format must sign to anonymize the data
(c) Software components and interfaces
Force10/Dell switches now. Alcatel-Lucent routers later this year.
(d)Experiences
We are happy with it
(e)Samples
See relevant RFCs
(f) Licenses or patents
no
3. Format details
(a)Transport protocol
Sflow, Netflow
(b)Structure or specification
i. Format specification
See RFCs
ii. Availability of specification
Yes, RFCs
iii.Extending the format
No
iv.Validate syntax and semantics
v. Representation
binary
68
Page 68 / 95
(c) Type of data or threat
(d)Security aspects
i. Confidentiality and integrity
Content of data is confidential
ii. Authentication
no
iii.Availability
(e)User group
CERT, statisitics
(f) Communication infrastructure
i. Peer to peer
ii. Centralised
iii.Closed user group
Definitely yes – legal aspects apply
(g)Software components
Any sflow/netflow software like Arbor...
69
Page 69 / 95
8.11. Questionnaire J
1. Format name and version
HPFEEDS is not a data format, but a transport protocol over TCP used to convey honeypots
data feeds.
More information can be found at https://redmine.honeynet.org/projects/hpfeeds/wiki
It is widely used and developed by honeynet project crew (http://www.honeynet.org/about)
2. Use case
We use HPFEEDS protocol to collect data from heterogeneous honeypots belonging to our
honeynet.
It is currently supported by a variety of honeypots:
1. dionaea http://dionaea.carnivore.it/,
2. kippo https://code.google.com/p/kippo/
3. glastopf http://glastopf.org/
and also by cuckoo sandbox http://www.cuckoosandbox.org/
Any kind of data format can be carried by this protocol without any constraints.
(a)Role and rationale
The "hpfeeds" project implements a lightweight authenticated publish/subscribe protocol for
exchanging live datafeeds.
Different feeds are separated by channels and support arbitrary binary payloads. This means
that the channel users have to decide about the structure of data. This could for example be
done by choosing a serialization format.
It provides authentication of each subscriber/publisher over each channel and optionally the
protocol can be run on top of SSL/TLS.
(b)Workflows
The main component is the so called “broker” that collects and dispatches live feeds among
publishers and subscribers through authenticated channel. Each source can send and/or
receive information in real time by publishing and/or subscribing to different channels.
Data format carried by each channel is not defined by the protocol, but have to be previously
set by parties interested in the communication over the specific channels.
Nowadays most of existing channels uses JSON (http://www.json.org/) to exchange data
(c) Software components and interfaces
We use available implementation of broker and publisher (honeypot plugin/patches)
component provided by honeynet project team (https://github.com/rep/hpfeeds).
Beside this we implemented proprietary software to parse and collect data (subscribers).
For debugging purposes Wireshark dissector has been implemented and included from latest
Wireshark release.
70
Page 70 / 95
(d)Experiences
Very simple and flexible protocol, easy to set up and operate.
Scalability of the solution should be analysed/improved. Now there is one central point (the
broker) that receives and relays all the messages. This should be a bottleneck and single
point of failure in a big deployment.
(e)Samples
For demonstration purposes, some example messages analysed with wireshark:
(f) Licenses or patents
HPFEEDS protocol is released under GNU PUBLIC LICENSE version 3
https://github.com/rep/hpfeeds/blob/master/LICENSE
3. Format details
HPFEEDS is not a data format specifications so many of the following questions do not apply
in this context.
Any kind of data format can be carried by this protocol without any constraints.
(a)Transport protocol
The protocol is carried by TCP optionally the protocol can be run on top of SSL/TLS.
71
Page 71 / 95
(b)Structure or specification
i. Format specification
ii. Availability of specification
iii.Extending the format
iv.Validate syntax and semantics
v. Representation
(c) Type of data or threat
(d)Security aspects
i. Confidentiality and integrity
Available only if protocol is run on top of SSL/TLS.
ii. Authentication
Currently supported
iii.Availability
(e)User group
(f) Communication infrastructure
i. Peer to peer
ii. Centralised
This is the way the protocol works as the central node is the broker component.
iii.Closed user group
(g)Software components
Protocol implementation is available at https://github.com/rep/hpfeeds
Most channels use JSON http://www.json.org as data format.
72
Page 72 / 95
8.12. Questionnaire K
1. Format name and version
IODEF (Incident Object Description Exchange Format) RFC 5070
2. Use case
(a)Role and rationale
CyDef receives some data in IODEF format from other parties, and also exports some data in
this format. The biggest factor for using IODEF is that it’s fairly simple and tailor-made for
exchanging incident reports with CSIRTs.
(b)Workflows
CyDef doesn’t store any data in IODEF format, but only converts to and from when
exchanging data with other response teams.
(c) Software components and interfaces
Bespoke parsing library.
(d)Experiences
IODEF works very well when exchanging blacklistings and similar data. Through XML
extensibility, it is able to include other events, such as attack patterns, vulnerabilities etc.
However, our opinion is that when using extended data types, STIX offers a more promising
solution (although we are yet to use it).
For this reason, we would recommend either using STIX for exchanging blacklisting data
(pro: more standardisation; con: quite bloated for simple blacklisting data), or using STIX for
the majority of data types with one or two exceptions, such as for blacklisting data.
(e)Samples
Samples available in RFC 5070: http://tools.ietf.org/html/rfc5070#section-7
(f) Licenses or patents
Rights are retained by the data owners. For full details, see IETF BCP 78 and IETF BCP 79.
3. Format details
(a)Transport protocol
No. Any protocol meeting certain requirements (confidentiality, integrity, authenticity, suitable
compression & reliability) is suitable.
(b)Structure or specification
i. Format specification
Yes.
ii. Availability of specification
Publicly available on IETF’s website. RFC 5070 covers the core specification, with others for
extensions (e.g. RFC 5901 for phishing).
73
Page 73 / 95
iii.Extending the format
Yes, but not realistic or advisable.
iv.Validate syntax and semantics
Yes, but no scripts are provided by IETF for this purpose. Bespoke code according to the
specification is required.
v. Representation
XML.
(c) Type of data or threat
Principally for exchanging blacklists and other incident reports. But also contains other
extensions (for phishing, attack patterns, vulnerabilities etc.).
(d)Security aspects
i. Confidentiality and integrity
Both are left to the transport protocol (it is not tied to a specific protocol).
ii. Authentication
Left to the transport protocol.
iii.Availability
Not covered by the data format and not possible to be covered by the transport protocol.
(e)User group
It is targeted towards CSIRTs, but has also been widely-used internally by corporations.
(f) Communication infrastructure
i. Peer to peer
ii. Centralised
iii.Closed user group
Preferred, as it was designed to be exchanged with full knowledge between individual
parties.
(g)Software components
No official software tools. However, several have been publicly released by CERTs and
CSIRTs.
74
Page 74 / 95
8.13. Questionnaire L
1. Format name and version
We are using IODEF format to exchange data on detected intrusion between SIEM systems
and our cyber security hypervisor.
It is possible to reuse this format to exchange malware information from our malware
analyzer to the security hypervisor.
2. Use case
IODEF data from customer networks to SOC hypervisor
(a)Role and rationale
Role of CSD (Cassidian CyberSecurity): security management
Role of customer: target of intrusions
Format used to describe event flows (src/tgt), nature of the incident, date of occurrence,
impact assessment
(b)Workflows
The SOC team manages different customers at the same time. Gathered Incident data are
imported by CSD from these customers. Depending on the customer, the incident resolution
is assigned to a specific team.
(c) Software components and interfaces
Interface between SIEM and hypervisor is IODEF\SOAP
(d)Experiences
IODEF is a very detailed format but most of the information is not used by SIEM systems
A lightest exchange format would be preferable. Sometimes we use syslog interfaces when
SIEM product has little information to transmit to the hypervisor.
(e)Samples
(f) Licenses or patents
SIEM are commercial products. Hypervisor is property of CSD.
IODEF\SOAP interface may be reused as web service (wsdl available).
3. Format details
(a)Transport protocol
IODEF\SOAP is an Http request
(b)Structure or specification
i. Format specification
WSDL which is an XML description of the requests and responses supported by the web
service
75
Page 75 / 95
ii. Availability of specification
WSDL can be delivered by CSD, a document describing the interface is also available
explaining the purpose and structure of the requests (function calls) & responses (function
returns)
iii.Extending the format
It is possible to add functions and/or to add parameters to existing functions
iv.Validate syntax and semantics
Yes it is: use of IODEF xsd
v. Representation
XML
(c) Type of data or threat
Designed for cyber security incidents
(d)Security aspects
i. Confidentiality and integrity
Both
ii. Authentication
Yes
iii.Availability
The format leverages the robustness of the HTTP protocol
(e)User group
(f) Communication infrastructure
i. Peer to peer
ii. Centralised
Preferred
iii.Closed user group
(g)Software components
Software applications used in this kind of exchanges are commercial products.
76
Page 76 / 95
8.14. Questionnaire M
1. Format name and version
Name is X-ARF, version of the specification is v0.2.
2. Use case
X-ARF data exchange between CSIRTs
(a)Role and rationale
DFN-CERT uses the X-ARF format to report incidents to other CSIRTs. The key advantage of
the format is the flexibility. X-ARF contains both a textual human readable as well as
structured part. The textual part can be understood without knowledge of the format and is
therefore intended for sites that are not used to X-ARF reports. However, the format allows
other sites to automate the processing of the reports.
(b)Workflows
The incident data is imported by DFN-CERT from different sources. Workflows exist to assign
the source of an incident to the appropriate site or CSIRT. The data is then used to produce
an X-ARF report that is sent to the site.
(c) Software components and interfaces
DFN-CERT uses an internal implementation of interfaces to import, parse, and export X-ARF
messages. Additionally, a sample script exists that inspects SSH server logs for attacks and
produces X-ARF reports.
(d)Experiences
X-ARF performs well for manual and automatic processing. A drawback is the inefficient data
transport when a separate mail for each event is transferred. This is especially true for bulk
data. To overcome this, an extension is part of the standard that provides a specification to
optionally aggregate multiple incidents in a single message. Moreover, a compression of the
textual data on the transport would lead to a further improvement of its efficiency. In the
current specification, X-ARF messages are transferred by email. Future releases may
consider a transport channel by HTTP (e.g. HTTP REST interface).
(e)Samples
It is attached below.
(f) Licenses or patents
It is not bound to any licenses or patents.
3. Format details
(a)Transport protocol
X-ARF messages are transferred by email (SMTP). Future specifications may also consider
the transport by HTTP/REST.
77
Page 77 / 95
(b)Structure or specification
i. Format specification
X-ARF messages are separated into three parts. The first is a textual description of the
content. The second part consists of a machine-readable part. Its structure is provided by
YAML/JSON. The third part is optional and may contain evidence of the incident (e.g. logs) or
malware samples.
ii. Availability of specification
Yes, at http://x-arf.org
iii.Extending the format
Yes, the specification includes a private schema. Additionally, other schemas regarding other
attack data can be proposed in collaboration with the working group.
iv.Validate syntax and semantics
Yes, this is true for the second part (validation of correct syntax)
v. Representation
All parts contain textual data.
(c) Type of data or threat
X-ARF provides multiple schemas related to different attack data. Schemas exist for portscanning activity, spam, and malware.
(d)Security aspects
i. Confidentiality and integrity
Yes, by using S/MIME signatures and encryption
ii. Authentication
Yes, by using S/MIME signatures
iii.Availability
The format leverages the robustness of the SMTP protocol.
(e)User group
X-ARF addresses different user groups. The first informal part is intended for users that are
not familiar with X-ARF while the second part is machine-readable and supports automation.
(f) Communication infrastructure
i. Peer to peer
ii. Centralised
iii.Closed user group
X-ARF supports all communication infrastructures.
78
Page 78 / 95
(g)Software components
The software is available at http://x-arf.org. It is not bound to any licenses or patents.
Sample of X-ARF message
From: [email protected]
To: [email protected]
Reply-To: [email protected]
X-Data-Format: X-ARF
Organisation: xxxxxxxx
X-System-Id: xxxxx.xxxxxxxx.de
X-Script-Version: 2010-12-21
X-Script-Name: xarf-ssh-reporter.sh
X-ARF: yes
Auto-Submitted: auto-generated
Subject: abuse report about xxx.xxx.129.56 - 2012-06-10
Mime-Version: 1.0
Content-Type: multipart/mixed; charset=utf8; boundary="Abuse64a4e26a2f19ad1616aa764f5edf8679"
Message-Id: <[email protected]>
Date: Sun, 10 Jun 2012 08:00:31 +0200 (CEST)
This message is in MIME format. But if you can see this,
you aren't using a MIME aware mail program. You shouldn't
have too many problems because this message is entirely in
ASCII and is designed to be somewhat readable with old
mail software.
--Abuse-64a4e26a2f19ad1616aa764f5edf8679
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Type: text/plain; charset=utf8;
Dear DFN-CERT,
this is an automated report for ip address xxx.xxx.129.56 in format "X-ARF"
generated on 2012-06-10 08:00:31 +0200
ip address xxx.xxx.129.56 produced 314 log lines, sample log lines attached.
Regards,
DFN-CERT Team
--Abuse-64a4e26a2f19ad1616aa764f5edf8679
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Type: text/plain; charset=utf8; name="report.txt";
--Reported-From: [email protected]
Category: abuse
Report-Type: login-attack
Service: ssh
Port: 22
User-Agent: xarf-ssh-reporter.sh 2010-12-21
Report-ID: [email protected]
Date: Sat, 09 Jun 2012 10:00:49 +0200
Source: xxx.xxx.129.56
Source-Type: ipv4
Attachment: text/plain
79
Page 79 / 95
Schema-URL: http://www.x-arf.org/schema/abuse_login-attack_0.1.1.json
--Abuse-64a4e26a2f19ad1616aa764f5edf8679
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Type: text/plain; charset=utf8; name="logfile.log";
2012-06-09 10:00:49 +0200 XXXXXX
string from xxx.xxx.129.56
2012-06-09 10:05:40 +0200 XXXXXX
xxx.xxx.129.56
2012-06-09 10:05:47 +0200 XXXXXX
xxx.xxx.129.56
-- MARK -2012-06-09 10:42:30 +0200 XXXXXX
xxx.xxx.129.56
2012-06-09 10:42:41 +0200 XXXXXX
xxx.xxx.129.56
2012-06-09 10:42:49 +0200 XXXXXX
xxx.xxx.129.56
sshd[26790]: Did not receive identification
sshd[27285]: Invalid user abdulghaffar from
sshd[27305]: Invalid user abdulkader from
sshd[970]: Invalid user atmail from
sshd[1000]: Invalid user atn from
sshd[1023]: Invalid user atowar from
--Abuse-64a4e26a2f19ad1616aa764f5edf8679--
80
Page 80 / 95
8.15. Questionnaire N
1. Format name and version
IDMEF
2. Use case
(a)Role and rationale
The specific use case of IDMEF is the transport of IDS data such as Snort to a central
storage centre. For example, the CarmentiS early warning system is capable of processing
IDMEF reports.
(b)Workflows
The primary purpose of IDMEF is enabling transportation of attack data from a distributed
network of IDS sensors.
(c) Software components and interfaces
NIDS such as Snort and Prelude export data. In addition, the Prelude framework provides a
programming library to produce, process, and import IDMEF data. The Prelude library is also
part of CarmentiS to process data.
(d)Experiences
The formats work pretty well for NIDS data. A nice feature is its capability to aggregate
multiple correlated events.
(e)Samples
Samples are provided in RFC4765
(f) Licenses or patents
No
3. Format details
(a)Transport protocol
No. Since the format is based on XML all protocols can be used that support XML.
(b)Structure or specification
i. Format specification
Yes, it is structured by XML; see RFC4765 for further details.
ii. Availability of specification
Yes, it is detailed in RFC4765
iii.Extending the format
Yes, IDMEF provides some means to extend the format.
81
Page 81 / 95
iv.Validate syntax and semantics
Yes
v. Representation
Representation is textual. However, some programs such as Prelude provide a binary
representation of IDMEF data.
(c) Type of data or threat
The format is devoted to IDS alerts.
(d)Security aspects
i. Confidentiality and integrity
Yes, e.g. by the Prelude library.
ii. Authentication
Yes, e.g. by the Prelude library.
iii.Availability
No
(e)User group
Since the format is intended to submit IDS data in an automated way, it is not addressed to a
specific user group.
(f) Communication infrastructure
i. Peer-to-peer
Yes
ii. Centralised
Yes
iii.Closed user group
No
(g)Software components
For example, the Prelude framework provides a free version of a library to process IDMEF
messages. It is published under the terms of the GNU General Public License.
82
Page 82 / 95
8.16. Questionnaire O
1. Format name and version
STIX (Structured Threat Information eXpression) V1.0
2. Use case
(a)Role and rationale
STIX (http://stix.mitre.org/) is a community driven effort to develop a standardized threat
information format. It is coordinated by Mitre, and as such it extends work on previous
standards they have produced, with a STIX message potentially including Cyber
Observables (CybOX), Malware Definitions (MAEC) and Attack Patterns (CAPEC). STIX
combines structured XML that describes observed security related events and artefacts with
a framework that caters for analysis elements.
There is a great deal of interest in using STIX, as it appears to offer high functionality in a
well defined standard that will facilitate automated exchange of threat information. LSEC, the
ACDC WP2 leaders, are therefore initiating a small project to examine how STIX could be
utilised as a data format for tool reporting in cooperation with a small number of ACDC tool
providers.
(b)Workflows
It is intended that information logged or otherwise provided by tools will be either directly
created as STIX messages, or converted from their current format into STIX. The STIX
messages will then be stored centrally, where the benefits of a common format reported by
disparate tools can be examined.
(c) Software components and interfaces
A STIX demonstrator system will be produced, which will provide a centralised, database
backed store, with a web service interface that allows authorized tools to submit data in STIX
format. The web service may be an implementation of the Trusted Automated eXchange of
Indicator Information (TAXII) protocol with XML binding or a simpler interface depending upon
ease of integration. The STIX demonstrator will be a STIX consumer, tools that send STIX
information will be STIX producers. STIX producers can be integrated directly with the STIX
demonstrator, or a command line stub will be made available that will allow easy integration
into existing reporting mechanisms without significant work by the tool partner.
(d)Experiences
At the moment this work has only just started, however the full intention is to provide
extensive feedback to the rest of the ACDC project.
(e)Samples
A
number
of
examples
of
STIX
documents
can
be
found
at
https://github.com/STIXProject/schemas/tree/master/samples however one example is also
included at the end of the questionnaire.
(f) Licenses or patents
The copyright for STIX and all associated Mitre initiatives belongs to the Mitre Corporation,
who openly grant a royalty free license for use (http://stix.mitre.org/about/termsofuse.html).
83
Page 83 / 95
3. Format details
(a)Transport protocol
STIX messages are well formed XML documents, and could be transported using many
Internet protocols, however there is a specific transport specification called TAXII which
includes bindings to HTTP and to XML for web services.
(b)Structure or specification
i. Format specification
Yes, STIX and all included data formats are defined by schema files maintained by Mitre on
behalf of the community.
ii. Availability of specification
Yes,
the
current
versions
of
http://stix.mitre.org/language/version1.0/
the
schema
files
are
available
here:
iii.Extending the format
From the website: “STIX also offers a set of loosely coupled schema extension points and
related default extensions for various purposes, such as use of externally-defined schemas
for relevant information, data marking models and controlled vocabularies.”.
iv.Validate syntax and semantics
Yes, by validating a STIX xml message against the schema files.
v. Representation
Messages are represented as well formed XML documents.
(c) Type of data or threat
STIX provides multiple schemas for representing various types threats and actors, including
Indicators, Threat Actors, Campaigns, Incidents, and Tactics, Techniques and Procedures
(TTP). The inclusion of elements from other standards such as CybOX allows many types of
observables to be included within the STIX document, such IP Addresses, E-mails,
Attachments, files, network packets etc.
(d)Security aspects
i. Confidentiality and integrity
If the STIX documents are transported via TAXII this could be done via TLS.
ii. Authentication
This is not part of STIX or TAXII, but would be the responsibility of the either the TAXII backend architecture, or security provided by the infrastructure, such as certificate authentication
on TLS.
iii.Availability
Nothing in the standards themselves.
84
Page 84 / 95
(e)User group
It is not aimed at any particular target group, but is intended to be able to widely support the
sharing of threat intelligence amongst interested parties. The standard has wide support
amongst CERTS, ISACs, commercial and government organisations in the US, ACDC could
be an early adopter in the EU.
(f) Communication infrastructure
i. Peer to peer
ii. Centralised
iii.Closed user group
TAXII describes “Hub and Spoke” which is a centralised approach, Full peer-to-peer, and
Source/Subscriber, which allows a source to push information to all subscribers.
(g)Software components
Python examples are supplied that demonstrate STIX document parsing
(https://github.com/STIXProject/python-stix), STIX is mainly about the definitions supplied by
the schema files, which can be readily handled using standard XML library tools in most
languages.
The following shows a STIX document that represents a threat indicator – in this case a list of
malicious URLs. The URLs are represented as a CybOX element within the STIX document.
<!--
STIX IP Watchlist Example
Copyright (c) 2013, The MITRE Corporation. All rights reserved.
The contents of this file are subject to the terms of the STIX License
located at http://stix.mitre.org/about/termsofuse.html.
This example demonstrates a simple usage of STIX to represent a list of URL
indicators (watchlist of URLs). Cyber operations and malware analysis centers
often share a list of suspected malicious URLs with information about what
those URLs might indicate. This STIX package represents a list of three URLs
addresses with a short dummy description of what they represent.
It demonstrates the use of:
*
*
*
*
*
STIX Indicators
CybOX within STIX
The CybOX URI Object (URL)
CybOX Patterns (apply_condition="ANY")
Controlled vocabularies
Created by Mark Davidson
-->
<stix:STIX_Package
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xmlns:stix="http://stix.mitre.org/stix-1"
xmlns:indicator="http://stix.mitre.org/Indicator-2"
xmlns:cybox="http://cybox.mitre.org/cybox-2"
xmlns:URIObject="http://cybox.mitre.org/objects#URIObject-2"
xmlns:cyboxVocabs="http://cybox.mitre.org/default_vocabularies-2"
xmlns:stixVocabs="http://stix.mitre.org/default_vocabularies-1"
xmlns:example="http://example.com/"
xsi:schemaLocation="
85
Page 85 / 95
http://stix.mitre.org/stix-1 ../stix_core.xsd
http://stix.mitre.org/Indicator-2 ../indicator.xsd
http://cybox.mitre.org/default_vocabularies-2
../cybox/cybox_default_vocabularies.xsd
http://stix.mitre.org/default_vocabularies-1
../stix_default_vocabularies.xsd
http://cybox.mitre.org/objects#URIObject-2
../cybox/objects/URI_Object.xsd"
>
<stix:STIX_Header>
<stix:Title>Example watchlist that contains URL
information.</stix:Title>
<stix:Package_Intent xsi:type="stixVocabs:PackageIntentVocab1.0">Indicators - Watchlist</stix:Package_Intent>
</stix:STIX_Header>
<stix:Indicators>
<stix:Indicator xsi:type="indicator:IndicatorType"
id="example:Indicator-db4a6ffe-61f0-488d-85a1-20bd5e360f37">
<indicator:Type xsi:type="stixVocabs:IndicatorTypeVocab-1.0" >URL
Watchlist</indicator:Type>
<indicator:Description>Sample URL Indicator for this
watchlist</indicator:Description>
<indicator:Observable id="example:Observable-Pattern-cc5c00ce98a6-4cbe-8474-59eaecdb018f">
<cybox:Object>
<cybox:Properties xsi:type="URIObject:URIObjectType">
<URIObject:Value condition="Equals"
apply_condition="ANY">http://example.com/foo/malicious1.html,http://example.co
m/foo/malicious2.html,http://example.com/foo/malicious3.html</URIObject:Value>
</cybox:Properties>
</cybox:Object>
</indicator:Observable>
</stix:Indicator>
</stix:Indicators>
</stix:STIX_Package>
86
Page 86 / 95
8.17. Questionnaire P
1. Format name and version
I+D (TID) tools data formats are:
• Spam-bot and DNS-bot detector, part of a DPI in house product, with real inline traffic.
Inputs and outputs are based on CSV formatted files for aggregate information. Planning
for a Standard data format like IODEF is scheduled.
• SDN Malware detector based on beta Commercial product use standard SYSLOG protocol
as an output.
ISP network haves his own data format use:
• Manual (e-mail & phone) information exchange with authorities, CERTs, ISPs, etc.... There
is no data format define at the moment.
• E-mail abuse-mailbox: [email protected] Support email with open text format with claims
related to SPAM and botnet activity. Only one requirement: needs, as probe of the offense,
the original SPAM offending mail with ALL mail headers.
• HTTP in web page (http://www.xxxxx.es/xxxxx ) for complaints and abuse from final users,
ISP clients or other ISP. Includes options for:
• ISP complainant name & contact email
• Complainant person identification
• IP origin of attack
• IP destination of attack
• Comments
• Log data of the complaint.
• Type of complaint (scanning, infringement of IPR, SPAM, DoS,..)
87
Page 87 / 95
Following details are from TID tools.
2. Use case
(a)Role and rationale
Spam-bot & DNS-bot detectors allow inspecting network traffic and detecting ISP users
infected with a botnet. This solution is expected to be deploy inside a ISP Networks (not at
this moment). CSV Format is human readable, allow easy conversion to other formats and
integration in Databases.
SDN Malware detector allow detecting infected employees by botnet inside a Enterprise.
Syslog protocol allow near time real incident alert.
(b)Workflows
SDN Malware detector tool generate output flows (syslog) of real time detections of infected
user.
Spam.bot & DNS–bot module in DPI generate and export aggregate files with detections
between a time of period ( default value is 15 minutes). Also as part of a workflow we are
planning reports generation.
These tools are source of detections and therefore can export the information to a
Centralized point. Inputs requirements can be updates of IPs and Domains from Centralized
point or third sources to increase number of detections and mitigations.
(c) Software components and interfaces
SDN malware detector is based on Hardware switches with Openflow support, and a Virtual
Software OpenFlow Controller of a Third Vendor Beta product able to receive DNS traffic and
checks domains against several domains Blacklist. Positive detections generate syslog
messages.
Spam-bot & DNS-bot are proprietary software analysis module running in a Linux system.
Analysis is done with the information received from a generic HW DPI over proprietary
format. Python scripting libraries for integration with new data formats are preferred.
(d)Experiences
Live pilot experience with SDN Malware detector in TID network show that Syslog protocol
needs syslog servers infrastructure but also that generate accurate information ( real time
botnets activity). Perfect for centralized solutions like ACDC.
We would like to improve reports capacity generation in both tools.
We are planning a extension to new data formats. We are willing to have a common
reference to develop inside of the ACDC project.
(e)Samples
Output CSV file sample from Spam-bot detector DPI module:
ANALYSIS DATE:
1363507205
1363507242 1.1.1.1
1363507223 100.100.100.100
1363507493 10.10.10.10
0
0
0
166
160
149
0
0
0
0
0
0
0
0
0
388
352
389
MOBILE
LANDLINE
MOBILE
11454
11475
10723
88
Page 88 / 95
Input file of Botnet Domains for DNS-bot detector DPI module:
Malwarefamily.dbl:
99-300.ru
360safeupdate02.gicp.net
3apa3a.tomsk.tw
Syslog message from SDN Malware detector:
Mon May 27 13:03:11 CEST 2013 CEF:0|Vendor|Controller|1.0.0|55|DNS query
notification|6|msg=OF Switch ID: 00:00:00:00:00:00:00:00 InPort: 25 Score: 80,
Tags: Botnet dvc=10.0.1.1 src=10.1.2.138 act=DROP_NOTIFY dhost=malware.domain
Wed May 29 15:28:38 CEST 2013 CEF:0|Vendor|Controller|1.0.0|55|DNS query
notification|6|msg=OF Switch ID: 00:00:00:00:00:00:00:00 InPort: 27 Score: 0,
Tags: Custom blacklist (Web app) dvc=10.0.1.1 src=10.1.2.138 act=NOTIFY
dhost=custom_malware.domain
(f) Licenses or patents
SDN Malware Detector is based in beta testing Commercial product with license cost. Syslog
content message format is proprietary.
Spam-bot & DNS-bot detector module is based on proprietary and patented protocols of
developed DPI. Output formats can be standard formats. Licenses model is being studied.
3. Format details
(a)Transport protocol
SDN Malware Detector uses UDP/514 Syslog protocol.
Spam-bot & DNS-bot detector module doesn’t have any transport protocol requirement.
(b)Structure or specification
i. Format specification
SDN Malware Detector uses syslog message. Field separators “|”
<Datetime> CEF:<number>|<Vendor name>|<SDN Controller hostname>|<Version>|
<number>|DNS query notification|6|msg=OF Switch ID: <MAC Address> InPort:
<port number> Score: <value>, Tags:<type of domain> dvc=<switch_IP>
src=<infected IP> act=<NOTIFY,DROP,DROP_NOTIFY> dhost=<malware domain>
Spam-bot Detector output CSV:
First Line: ANALYSIS DATE:
<datetime_decimal>
Each following line fields separator (tab):
• Datetime decimal format when spammer wast first seen.
• Spammer IP address (public or Private)
• Detection trigger (Zero if no detection happen): number of sent mails
89
Page 89 / 95
• Detection trigger (Zero if no detection happen): DNS Queries
• Detection trigger (Zero if no detection happen): SMTP error response
• Detection trigger (Zero if no detection happen): number of different senders
• Detection trigger (Zero if no detection happen): SMTP sents.
• Network VLANs number
• Network access ( landline or mobile)
• Bytes consumed
ii. Availability of specification
No.
iii.Extending the format
Yes can be extended for CSV format as an evolving product in testing phase.
iv.Validate syntax and semantics
Could be done, but there is no available tools at the moment.
v. Representation
Human readable text in all case.
(c) Type of data or threat
SDN Malware Detector generates a atomic syslog alert from a user accessing to a malicious
domain. These domains are related with botnet controller, droppers, phising, etc. The data
include the user IP, domains accessed and actions done (drop, alert or both).
Spam-bot & DNS-bot Detector output CSV format are designed to collect identification of
infected users of spam botnets or generic botnets detected by SMTP and DNS protocols that
allow a ISP to remediate his users.
(d)Security aspects
i. Confidentiality and integrity
None mechanism is available at this development stage. There can be delegated to
standards protocols mechanism (like FTP, SCP, HTTP and so forth).
ii. Authentication
None mechanism is available at this development stage. Transport can be delegated to
standards protocols authentication mechanism (like FTP, HTTP and so forth).
iii.Availability
None mechanism is available at this development stage. Transport can be delegated to
standards protocols authentication mechanism (like FTP, HTTP and so forth).
90
Page 90 / 95
(e)User group
Target user group for Spam-bot & DNS-bot detector are all ISP clients (landline and mobile),
mainly Security Operations Centers (SOC) or TAC.
SDN Malware detector target user group is SME and Corporations companies.
(f) Communication infrastructure
i. Peer to peer
ii. Centralised
iii.Closed user group
Centralized communications or Closed user groups is preferred in most case because of
aggregation of data needed.
(g)Software components
Spam-bot & DNS-bot detector use DPI proprietary and patented software over general PC
architecture hardware, for data capture and aggregation. Python is used in import, export of
data information.
Proprietary Switch OS with standard Openflow protocol support and proprietary Java based
Controller is used for SDN Malware detector. Linux standard syslog daemon is used for
export information.
91
Page 91 / 95
8.18. Questionnaire Q
1. Format name and version
• Raw Data Events
Raw Data events are un-processed events, and the input that AHPS will accept from other
ACDC components.
Raw data collected from the Event Sources is received by AHPS connectors and stored in
raw data files. Each raw data event is represented as a single line in a raw data file, in the
form of a JSON object with a predefined format.
• AHPS Events
AHPS events are processed events, and the output that AHPS will generate in the context of
ACDC.
The Atos High Performance Security (AHPS) Event Format is based on the Distributed Audit
Services (XDAS)1 standard.
2. Use case
(a)Role and rationale
The AHPS would receive information (Raw data events) from other ACDC components, such
as network or device sensor tools. Once we know the source of the information and the
format, we will need to develop a Collector component and a Connector for each one, and
configure AHPS to use it.
The AHPS is mainly an analysis component that receives information from other sources and
normalizes, filters, correlates and analyses the information received to automatically identify
inconsistencies in the environment. Based on these inconsistencies, AHPS identifies and
alert on anomalous activity or new suspicious trends, alerting of potential threats or attacks.
The information generated by AHPS (AHPS events) could be used as input to other ACDC
tools and stored in the CCH for further analysis.
(b)Workflows
Event Source
AHPS Connector
AHPS Collector
AHPS Core (backend)
Raw Data Event
AHPS Solution
AHPS storage
AHPS Integrator
AHPS Integrator
AHPS Event
ACDC component
AHPS Integrator
AHPS Event
ACDC component
ACDC component
1Distributed Audit Service (XDAS) https://www2.opengroup.org/ogsys/catalog/p441
92
Page 92 / 95
The AHPS takes the input from event source “connectors” and converts the raw data into a
textual map form consumable by, what is called “collectors”. Collectors parse and normalize
the textual map and create an AHPS Event, categorizing it according to the AHPS taxonomy
of events. The AHPS Event is enriched with additional source-specific data and, depending
on the collector, may apply additional contextual metadata such as identity, host, vulnerability,
or custom mapped metadata. AHPS events can be sent to other external systems or
components by means of “Integrators”.
It is possible to export or download the raw data files containing raw data events, collected
from event sources and stored in AHPS, in the CSV format.
AHPS Events will be output of AHPS and can be imported and exported from the internal
AHPS database through the AHPS web interface.
(c) Software components and interfaces
• Software: AHPS solution
• Interfaces:
• AHPS web interface: user interface for configuration and interaction with the solution
• Event Source Connector API: Java API
• Event Collectors API: JSP API
• Report generation interface
• Integrator API
(d)Experiences
(e)Samples
(f) Licenses or patents
Yes, both the AHPS Event Format and Raw data format are Atos proprietary.
3. Format details
(a)Transport protocol
No, it is a matter of configuration. AHPS can collect data from a wide range of event sources,
such as intrusion detection systems, firewalls, operating systems, routers, databases,
switches, mainframes, antivirus applications, and other applications. The configuration
required to integrate a new event source with AHPS varies, depending on the type of event
source and the communication method selected. For example, to accept Syslog data from
Syslog event sources that send data over TCP (port 1468),UDP(port 1514), or SSL(port
1443). You can also configure AHPS to listen on additional ports.
(b)Structure or specification
i. Format specification
Yes, for both Raw data events and AHPS events
ii. Availability of specification
No, they are not publicly available. The AHPS event format is based on XDAS standard.
93
Page 93 / 95
iii.Extending the format
The AHPS event format allows for custom extensions by using extension fields.
The Raw data event format is not extensible, but it is possible to create “mappers” to
transform from one origin format to the AHPS raw data format, and there is a field that works
as a payload, where the original event can be dumped.
iv.Validate syntax and semantics
Yes.
(c) Representation
Textual.
(d)Type of data or threat
AHPS raw data event format is used to describe events collected from the following sources:
• Security Perimeter: Devices and software used to create a security parameter for your
environment.
• Operating Systems: Events from the different operating systems running in the network.
• Referential IT Sources: The software used to maintain and track assets, patches,
configuration, and vulnerability.
• Application Events: Events generated from the applications installed in the network.
• User Access Control: Events generated from applications or devices that allow users
access to company resources.
The AHPS Event model describes event activity generated from integrated devices, services,
and applications. The fields of the event may describe a complex resource such as a user
account residing in a directory hosted by a particular server, or software module running
inside a service hosted by a particular server, and so forth. The AHPS Events are classified
according to a taxonomy, which uses the XDAS standard taxonomy (v1), a classification that
is intended to group events of similar type together to ease reporting and searching. Rather
than use proprietary, app-specific event names (login, authenticated, logged in, etc), all
events of a particular type should map to the same taxonomic classification.
AHPS Events are correlated and analyzed by AHPS to provide notifications about incidents
and attacks. Also, there is a feature to cross‐reference between event data signatures and
vulnerability scanner data, generating feeds which contain information about vulnerabilities
and threats, and associated remediation information.
(e)Security aspects
i. Confidentiality and integrity
Raw data can be checked for integrity by using the corresponding AHPS UI option. This
feature checks integrity by various means, for example:
• Verify the sequence number of JSON records, by using the fields ChainID and
ChainSequrence
• Verify the RawDataHash against the RawData
94
Page 94 / 95
ii. Authentication
Secured data collection is determined by the specific protocols supported by the event
source.
Internally, the protocol used for communication between the server and the database is
defined by a JDBC Driver. For networked storage locations to store the event data and raw
data, it depends on the capabilities of the type of server used. For example, CIFS or NFS
servers do not offer data encryption, while local or SAN storage servers do not have the
same security vulnerabilities.
AHPS uses several digital, public‐key certificates as part of establishing secure TLS/SSL
communications.
iii.Availability
(f) User group
Target group: Enterprises
(g)Communication infrastructure
The preferred communication infrastructure would be AHPS to run as a external service for
ACDC solution, which will receive events from other AHPS components (e.g. network
sensors, vulnerability scanners, etc) and will produce as output AHPS events (representing
attacks, incidents, threats, vulnerabilities, etc.) as well as reports and countermeasures.
i. Peer-to-peer
ii. Centralised
iii.Closed user group
(h)Software components
AHPS provides interfaces for import, export, parse, mapping, normalization, correlation data,
but all of them are proprietary. However, it is possible to develop components to adapt to the
specific formats of external components.
Statement of originality:
This deliverable contains original unpublished work except where
clearly indicated otherwise. Acknowledgement of previously published
material and of the work of others has been made through appropriate
citation, quotation or both.
95
Page 95 / 95